Context
In May 2026, we released SmithDB, a new observability database built for modern AI agents. SmithDB delivers industry-leading performance across every key observability workload, making core LangSmith experiences dramatically faster. SmithDB-powered methods are now available to SDK users in eligible regions.Deployment support
| Deployment | Status |
|---|---|
| US SaaS | Available. Methods are fully SmithDB-backed. |
| EU and other SaaS regions | Methods are available but not yet backed by SmithDB. |
| Self-hosted | Supported starting with self-hosted version 0.16.0. No feature flag is required. |
Minimum SDK version
The SmithDB-backed methods require a minimum SDK version:| Language | Package | Minimum version |
|---|---|---|
| Python | langsmith | >=0.10.2 |
| TypeScript | langsmith | >=0.8.1 |
| Java | langsmith-java | 0.1.0-beta.11 |
| Go | langsmith-go | v0.17.0 |
Migrate with an AI agent
This guide is written to be fetched and applied directly by an AI coding agent. Copy the following prompt into your agent to migrate your codebase to the SmithDB-backed methods.Migrate this codebase's LangSmith SDK usage to the new SmithDB-backed methods.
Fetch https://docs.langchain.com/langsmith/smithdb-sdk-migration.md and treat it
as the source of truth for what changed, including which methods and
parameters are affected, what replaces them, and deployment support.
1. Check the installed LangSmith SDK version against the minimum version
required for the SmithDB-backed methods per the guide, and upgrade the
dependency if it does not meet that minimum.
2. Identify every call site in this codebase that uses a method the guide
marks as migrated, in whichever language(s) this codebase uses.
3. For each call site, apply the corresponding before/after change from the
guide, including any added, removed, or renamed parameters.
If a call site or parameter is not covered by the guide, stop and ask rather
than guessing.
Exceptions
- Python
- TypeScript
- Java
- Go
- cURL
The SmithDB-backed methods raise new exception classes instead of the legacy
langsmith.utils exception classes.Before (langsmith.utils) | After (langsmith) | Notes |
|---|---|---|
LangSmithError | LangsmithError | Base exception class for the SDK; casing changed |
LangSmithAPIError | InternalServerError | 5xx |
LangSmithRequestTimeout | APITimeoutError | Raised when a request times out |
LangSmithUserError | (removed) | No direct equivalent. The 403 org-scoped-key case now raises PermissionDeniedError; client-side argument validation now raises a standard ValueError or TypeError |
LangSmithRateLimitError | RateLimitError | 429; unchanged name |
LangSmithAuthError | AuthenticationError | 401 |
LangSmithNotFoundError | NotFoundError | 404; unchanged name |
LangSmithConflictError | ConflictError | 409; unchanged name |
LangSmithConnectionError | APIConnectionError | Raised when the client cannot connect to the API |
LangSmithExceptionGroup | (removed) | No equivalent |
| (not available) | APIError | New: base class for all API-related errors, with message, request, and body attributes |
| (not available) | APIStatusError | New: base class for all 4xx/5xx status errors |
| (not available) | BadRequestError | New: 400 |
| (not available) | PermissionDeniedError | New: 403 |
| (not available) | UnprocessableEntityError | New: 422 |
| (not available) | APIResponseValidationError | New: raised when a response does not match the expected schema |
The SmithDB-backed methods raise new exception classes instead of plain
Error.Before (plain Error) | After (langsmith) | Notes |
|---|---|---|
| (not available) | LangsmithError | base class for all SDK errors |
| (not available) | InternalServerError | 5xx |
| (not available) | APIConnectionTimeoutError | Raised when a request times out |
| (not available) | RateLimitError | 429 |
| (not available) | AuthenticationError | 401 |
| (not available) | NotFoundError | 404 |
| (not available) | ConflictError | 409 |
| (not available) | APIConnectionError | Raised when the client cannot connect to the API |
| (not available) | APIError | base class for all API-related errors, with status, headers, and error properties |
| (not available) | BadRequestError | 400 |
| (not available) | PermissionDeniedError | 403 |
| (not available) | UnprocessableEntityError | 422 |
| (not available) | APIUserAbortError | Raised when a request is aborted via an AbortController |
No change. Error handling is unaffected by this migration.
No change. Error handling is unaffected by this migration.
No change. Error handling is unaffected by this migration.
Runs: query
Query runs from a project with optional filtering and field projection. Returns a paginated result set.Main changes
Method name
- Python
- TypeScript
- Java
- Go
- cURL
| Before | After |
|---|---|
client.list_runs() | client.runs.query() |
client.runs.query() is now async. Call it with await.| Before | After |
|---|---|
client.listRuns() | client.runs.query() |
| Before | After |
|---|---|
client.runs().query() | client.runs().queryV2() |
| Before | After |
|---|---|
client.Runs.Query() | client.Runs.QueryV2() |
| Before | After |
|---|---|
POST /api/v1/runs/query | POST /v2/runs/query |
Query parameters
- Python
- TypeScript
- Java
- Go
- cURL
project_name is not supported in runs.query. Pass project_ids with the project UUID instead. To look up a UUID by name, use client.read_project(project_name="my-project"), or await client.aread_project(project_name="my-project") in async code.min_start_time defaults to 1 day ago when omitted. list_runs with no start_time returned all historical runs; runs.query without min_start_time silently scopes the query to the last 24 hours. Pass an explicit min_start_time if you need a wider window.Before (list_runs) | After (runs.query) | Notes |
|---|---|---|
project_name | (removed) | Use project_ids with UUID(s)—see warning above |
project_id | project_ids | Now takes a list; mutually exclusive with reference_dataset_id |
run_type | run_type | Values must now be uppercase: "LLM", "CHAIN", "TOOL", "RETRIEVER", "EMBEDDING", "PROMPT", "PARSER" |
trace_id | trace_id | Unchanged |
reference_example_id | reference_examples | Now takes a list of UUIDs |
query | (removed) | No equivalent |
filter | filter | Syntax unchanged |
trace_filter | trace_filter | Unchanged |
tree_filter | tree_filter | Unchanged |
is_root | is_root | Unchanged |
parent_run_id | (removed) | No equivalent |
start_time | min_start_time | Renamed; defaults to 1 day ago—see warning above |
error | has_error | Renamed |
run_ids | ids | Renamed |
select | selects | Field names are now uppercase ("NAME", "STATUS", etc.) |
limit | (removed) | Use page_size for per-request batch size |
| (not available) | max_start_time | Upper bound for start_time; defaults to now |
| (not available) | page_size | Per-request result count (default 100, max 1000) |
| (not available) | reference_dataset_id | Alternative to project_ids; mutually exclusive |
| (not available) | cursor | Pass next_cursor from previous response to fetch next page |
projectName is not supported in client.runs.query. Pass project_ids with the project UUID instead. To look up a UUID by name, use client.readProject({ projectName: "my-project" }).min_start_time defaults to 1 day ago when omitted. listRuns with no startTime returned all historical runs; client.runs.query without min_start_time silently scopes the query to the last 24 hours. Pass an explicit min_start_time if you need a wider window.Before (listRuns) | After (client.runs.query) | Notes |
|---|---|---|
projectName | (removed) | Use project_ids with UUID(s)—see warning above |
projectId | project_ids | Renamed to snake_case; now takes a list; mutually exclusive with reference_dataset_id |
runType | run_type | Renamed to snake_case; values must now be uppercase: "LLM", "CHAIN", "TOOL", "RETRIEVER", "EMBEDDING", "PROMPT", "PARSER" |
traceId | trace_id | Renamed to snake_case |
referenceExampleId | reference_examples | Renamed to snake_case; now takes a list of UUIDs |
query | (removed) | No equivalent |
filter | filter | Syntax unchanged |
traceFilter | trace_filter | Renamed to snake_case |
treeFilter | tree_filter | Renamed to snake_case |
isRoot | is_root | Renamed to snake_case |
parentRunId | (removed) | No equivalent |
startTime | min_start_time | Renamed to snake_case; defaults to 1 day ago—see warning above |
error | has_error | Renamed |
id | ids | Renamed |
select | selects | Field names are now uppercase ("NAME", "STATUS", etc.) |
limit | (removed) | Use page_size for per-request batch size |
order | (removed) | No equivalent |
executionOrder | (removed) | No equivalent |
| (not available) | max_start_time | Upper bound for start_time; defaults to now |
| (not available) | page_size | Per-request result count (default 100, max 1000) |
| (not available) | reference_dataset_id | Alternative to project_ids; mutually exclusive |
| (not available) | cursor | Pass next_cursor from previous response to fetch next page |
minStartTime() defaults to 1 day ago when omitted. query() with no startTime() returned all historical runs; queryV2() without minStartTime() silently scopes the query to the last 24 hours. Pass an explicit minStartTime() if you need a wider window.Before (RunQueryParams) | After (RunQueryV2Params) | Notes |
|---|---|---|
session() | projectIds() | Renamed; now takes explicit project UUIDs |
runType() | runType() | Values must now be uppercase |
trace() | traceId() | Renamed |
referenceExample() | referenceExamples() | Renamed to plural |
query() | (removed) | No equivalent |
filter() | filter() | Syntax unchanged |
traceFilter() | traceFilter() | Unchanged |
treeFilter() | treeFilter() | Unchanged |
isRoot() | isRoot() | Unchanged |
parentRun() | (removed) | No equivalent |
startTime() | minStartTime() | Renamed; defaults to 1 day ago—see warning above |
error() | hasError() | Renamed |
id() | ids() | Renamed |
select() | selects() | Field names are now uppercase |
limit() | (removed) | Use pageSize() |
order() | (removed) | No equivalent |
executionOrder() | (removed) | No equivalent |
cursor() | cursor() | Unchanged |
| (not available) | maxStartTime() | Upper bound for start time; defaults to now |
| (not available) | pageSize() | Per-request result count (default 100, max 1000) |
| (not available) | referenceDatasetId() | Alternative to projectIds() |
MinStartTime defaults to 1 day ago when omitted. Query() with no StartTime returned all historical runs; QueryV2() without MinStartTime silently scopes the query to the last 24 hours. Pass an explicit MinStartTime if you need a wider window.Before (RunQueryParams) | After (RunQueryV2Params) | Notes |
|---|---|---|
Session | ProjectIDs | Renamed; now takes explicit project UUIDs |
RunType | RunType | Values must now be uppercase: RunQueryV2ParamsRunTypeLLM, RunQueryV2ParamsRunTypeChain, etc. |
Trace | TraceID | Renamed |
ReferenceExample | ReferenceExamples | Renamed to plural |
Query | (removed) | No equivalent |
Filter | Filter | Unchanged |
TraceFilter | TraceFilter | Unchanged |
TreeFilter | TreeFilter | Unchanged |
IsRoot | IsRoot | Unchanged |
ParentRun | (removed) | No equivalent |
StartTime | MinStartTime | Renamed; defaults to 1 day ago—see warning above |
Error | HasError | Renamed |
ID | IDs | Renamed |
Select | Selects | Field name constants are now uppercase (e.g., RunQueryV2ParamsSelectName) |
Limit | (removed) | Use PageSize |
Order | (removed) | No equivalent |
ExecutionOrder | (removed) | No equivalent |
Cursor | Cursor | Unchanged |
| (not available) | MaxStartTime | Upper bound for start time; defaults to now |
| (not available) | PageSize | Per-request result count (default 100, max 1000) |
| (not available) | ReferenceDatasetID | Alternative to ProjectIDs |
min_start_time defaults to 1 day ago when omitted. POST /api/v1/runs/query with no start_time returned all historical runs; POST /v2/runs/query without min_start_time silently scopes the query to the last 24 hours. Pass an explicit min_start_time if you need a wider window.Before (v1 POST /api/v1/runs/query body field) | After (v2 POST /v2/runs/query body field) | Notes |
|---|---|---|
session | project_ids | Renamed; both take an array of project UUIDs. project_ids is mutually exclusive with reference_dataset_id |
run_type | run_type | Values must now be uppercase: "LLM", "CHAIN", "TOOL", "RETRIEVER", "EMBEDDING", "PROMPT", "PARSER" |
trace | trace_id | Renamed |
reference_example | reference_examples | Renamed to plural; now takes an array of UUIDs |
query | (removed) | No equivalent |
filter | filter | Syntax unchanged |
trace_filter | trace_filter | Unchanged |
tree_filter | tree_filter | Unchanged |
is_root | is_root | Unchanged |
parent_run | (removed) | No equivalent |
start_time | min_start_time | Renamed; defaults to 1 day ago—see warning above |
error | has_error | Renamed |
id | ids | Renamed to plural |
select | selects | Field names are now uppercase ("NAME", "STATUS", etc.) |
limit | (removed) | Use page_size for per-request batch size |
| (not available) | max_start_time | Upper bound for start_time; defaults to now |
| (not available) | page_size | Per-request result count (default 100, max 1000) |
| (not available) | reference_dataset_id | Alternative to project_ids; mutually exclusive |
| (not available) | cursor | Pass next_cursor from previous response to fetch next page |
Response fields
- Python
- TypeScript
- Java
- Go
- cURL
Pass SCREAMING_SNAKE_CASE strings to
selects (eg. "ID", "NAME", "STATUS") to control which fields are populated on each Run; only selected fields are non-None. Default selects contains only "ID".Before (v1 Run attribute) | After (v2 Run attribute) | Notes |
|---|---|---|
run.id | run.id | Unchanged; returned by default when selects is omitted |
run.name | run.name | Unchanged |
run.run_type | run.run_type | Values are now uppercase Literals: "LLM", "CHAIN", etc. |
run.status | run.status | Values: "SUCCESS", "ERROR", "PENDING" |
run.start_time | run.start_time | Unchanged |
run.end_time | run.end_time | Unchanged |
run.error | run.error | Unchanged |
run.inputs | run.inputs | Unchanged |
run.outputs | run.outputs | Unchanged |
run.tags | run.tags | Unchanged |
run.extra | run.extra | Unchanged |
run.metadata | run.metadata | Unchanged |
run.events | run.events | Unchanged |
run.reference_example_id | run.reference_example_id | Unchanged |
run.trace_id | run.trace_id | Unchanged |
run.dotted_order | run.dotted_order | Unchanged |
run.parent_run_id | (removed) | Use run.parent_run_ids (list of all ancestor UUIDs, root first) |
run.parent_run_ids | run.parent_run_ids | Unchanged |
run.session_id | run.project_id | Renamed; session_id was the project UUID |
run.feedback_stats | run.feedback_stats | Unchanged |
run.app_path | run.app_path | Unchanged |
run.attachments | run.attachments | v2 returns pre-signed download URLs instead of raw bytes |
run.total_tokens | run.total_tokens | Unchanged |
run.prompt_tokens | run.prompt_tokens | Unchanged |
run.completion_tokens | run.completion_tokens | Unchanged |
run.total_cost | run.total_cost | Unchanged |
run.prompt_cost | run.prompt_cost | Unchanged |
run.completion_cost | run.completion_cost | Unchanged |
run.first_token_time | run.first_token_time | Unchanged |
run.latency (property) | run.latency_seconds | Renamed; was a computed timedelta property, now a native float field |
run.in_dataset | run.is_in_dataset | Renamed |
run.child_run_ids | (removed) | No equivalent |
run.child_runs | (removed) | No equivalent |
run.serialized | (removed) | Use run.manifest |
run.manifest_id | (removed) | Use run.manifest |
| (not available) | run.is_root | New |
| (not available) | run.manifest | New: full manifest object (replaces serialized and manifest_id) |
| (not available) | run.error_preview | New: truncated error snippet |
| (not available) | run.inputs_preview | New: truncated inputs preview |
| (not available) | run.outputs_preview | New: truncated outputs preview |
| (not available) | run.thread_id | New: conversation thread UUID |
| (not available) | run.reference_dataset_id | New: dataset UUID for the reference example |
| (not available) | run.share_url | New: public share URL (only set when the run has been shared) |
run.prompt_token_details | run.prompt_token_details.raw | Field now wraps the dict; access .raw to get dict[str, int] (element type unchanged) |
run.completion_token_details | run.completion_token_details.raw | Field now wraps the dict; access .raw to get dict[str, int] (element type unchanged) |
run.prompt_cost_details | run.prompt_cost_details.raw | Field now wraps the dict; access .raw to get dict[str, float] (was dict[str, Decimal]) |
run.completion_cost_details | run.completion_cost_details.raw | Field now wraps the dict; access .raw to get dict[str, float] (was dict[str, Decimal]) |
Pass SCREAMING_SNAKE_CASE strings to
selects (eg. "ID", "NAME", "STATUS") to control which fields are populated on each Run. Default selects contains only "ID".Before (v1 Run property) | After (v2 Run property) | Notes |
|---|---|---|
run.id | run.id | Unchanged |
run.name | run.name | Unchanged |
run.runType | run.run_type | Renamed to snake_case; values are now uppercase: "LLM", "CHAIN", etc. |
run.status | run.status | Values: "SUCCESS", "ERROR", "PENDING" |
run.startTime | run.start_time | Renamed to snake_case |
run.endTime | run.end_time | Renamed to snake_case |
run.error | run.error | Unchanged |
run.inputs | run.inputs | Unchanged |
run.outputs | run.outputs | Unchanged |
run.tags | run.tags | Unchanged |
run.extra | run.extra | Unchanged |
| (not available) | run.metadata | New: previously accessed via run.extra.metadata |
run.events | run.events | Unchanged |
run.referenceExampleId | run.reference_example_id | Renamed to snake_case |
run.traceId | run.trace_id | Renamed to snake_case |
run.dottedOrder | run.dotted_order | Renamed to snake_case |
run.parentRunId | (removed) | Use run.parent_run_ids (list of all ancestor UUIDs, root first) |
run.parentRunIds | run.parent_run_ids | Renamed to snake_case |
run.sessionId | run.project_id | Renamed; sessionId was the project UUID |
run.feedbackStats | run.feedback_stats | Renamed to snake_case |
run.appPath | run.app_path | Renamed to snake_case |
run.attachments | run.attachments | v2 returns pre-signed download URLs instead of raw bytes |
run.totalTokens | run.total_tokens | Renamed to snake_case |
run.promptTokens | run.prompt_tokens | Renamed to snake_case |
run.completionTokens | run.completion_tokens | Renamed to snake_case |
run.totalCost | run.total_cost | Renamed to snake_case |
run.promptCost | run.prompt_cost | Renamed to snake_case |
run.completionCost | run.completion_cost | Renamed to snake_case |
run.firstTokenTime | run.first_token_time | Renamed to snake_case |
run.latency | run.latency_seconds | Renamed; was a computed property, now a native number field (seconds) |
run.inDataset | run.is_in_dataset | Renamed |
run.childRunIds | (removed) | No equivalent |
run.childRuns | (removed) | No equivalent |
run.serialized | (removed) | Use run.manifest |
run.manifestId | (removed) | Use run.manifest |
run.shareToken | (removed) | Use run.share_url (full URL, only set when the run has been shared) |
| (not available) | run.is_root | New |
| (not available) | run.manifest | New: full manifest object (replaces serialized and manifestId) |
| (not available) | run.error_preview | New: truncated error snippet |
| (not available) | run.inputs_preview | New: truncated inputs preview |
| (not available) | run.outputs_preview | New: truncated outputs preview |
| (not available) | run.thread_id | New: conversation thread UUID |
| (not available) | run.reference_dataset_id | New: dataset UUID for the reference example |
| (not available) | run.share_url | New: public share URL (only set when the run has been shared) |
| (not available) | run.prompt_token_details | New: per-category prompt token breakdown |
| (not available) | run.completion_token_details | New: per-category completion token breakdown |
| (not available) | run.prompt_cost_details | New: per-category prompt cost breakdown |
| (not available) | run.completion_cost_details | New: per-category completion cost breakdown |
Add
RunQueryV2Params.Select values (eg. Select.NAME, Select.STATUS) via .addSelect(...) to control which fields are populated; unselected fields return empty Optional values. selects() defaults to ID only.Before (RunSchema method) | After (Run method) | Notes |
|---|---|---|
run.id() | run.id() | Unchanged |
run.name() | run.name() | Unchanged |
run.runType() | run.runType() | Values are now uppercase: "LLM", "CHAIN", etc. |
run.status() | run.status() | Values: "SUCCESS", "ERROR", "PENDING" |
run.startTime() | run.startTime() | Unchanged |
run.endTime() | run.endTime() | Unchanged |
run.error() | run.error() | Unchanged |
run.inputs() | run.inputs() | Unchanged |
run.outputs() | run.outputs() | Unchanged |
run.tags() | run.tags() | Unchanged |
run.extra() | run.extra() | Unchanged |
run.events() | run.events() | Unchanged |
run.feedbackStats() | run.feedbackStats() | Unchanged |
run.inputsPreview() | run.inputsPreview() | Unchanged |
run.outputsPreview() | run.outputsPreview() | Unchanged |
run.referenceExampleId() | run.referenceExampleId() | Unchanged |
run.traceId() | run.traceId() | Unchanged |
run.dottedOrder() | run.dottedOrder() | Unchanged |
run.parentRunId() | (removed) | Use run.parentRunIds() (list of all ancestor UUIDs, root first) |
run.parentRunIds() | run.parentRunIds() | Unchanged |
run.sessionId() | run.projectId() | Renamed; sessionId() returned the project UUID |
run.appPath() | run.appPath() | Unchanged |
run.firstTokenTime() | run.firstTokenTime() | Unchanged |
run.totalTokens() | run.totalTokens() | Unchanged |
run.promptTokens() | run.promptTokens() | Unchanged |
run.completionTokens() | run.completionTokens() | Unchanged |
run.totalCost() | run.totalCost() | Return type changed from Optional<String> to Optional<Double> |
run.promptCost() | run.promptCost() | Return type changed from Optional<String> to Optional<Double> |
run.completionCost() | run.completionCost() | Return type changed from Optional<String> to Optional<Double> |
run.promptTokenDetails() | run.promptTokenDetails() | Unchanged |
run.completionTokenDetails() | run.completionTokenDetails() | Unchanged |
run.promptCostDetails() | run.promptCostDetails() | Unchanged |
run.completionCostDetails() | run.completionCostDetails() | Unchanged |
run.priceModelId() | run.priceModelId() | Unchanged |
run.inDataset() | run.isInDataset() | Renamed |
run.referenceDatasetId() | run.referenceDatasetId() | Unchanged |
run.threadId() | run.threadId() | Unchanged |
run.shareToken() | (removed) | Use run.shareUrl() (full URL, only set when the run has been shared) |
run.childRunIds() | (removed) | No equivalent |
run.directChildRunIds() | (removed) | No equivalent |
run.serialized() | (removed) | Use run.manifest() |
run.manifestId() | (removed) | Use run.manifest() |
run.messages() | (removed) | No equivalent |
run.executionOrder() | (removed) | No equivalent |
run.lastQueuedAt() | (removed) | No equivalent |
run.traceFirstReceivedAt() | (removed) | No equivalent |
run.traceMaxStartTime() | (removed) | No equivalent |
run.traceMinStartTime() | (removed) | No equivalent |
run.traceTier() | (removed) | No equivalent |
run.traceUpgrade() | (removed) | No equivalent |
run.ttlSeconds() | (removed) | No equivalent |
| (not available) | run.attachments() | New: pre-signed download URLs for attachments (replaces S3 URL fields) |
| (not available) | run.latencySeconds() | New: wall-clock duration in seconds |
| (not available) | run.isRoot() | New |
| (not available) | run.errorPreview() | New: truncated error snippet |
| (not available) | run.manifest() | New: full manifest, typed as Optional<Manifest> (replaces serialized() and manifestId()) |
| (not available) | run.metadata() | New: metadata, typed as Optional<Metadata> (was derived from extra.metadata) |
| (not available) | run.shareUrl() | New: public share URL (only set when the run has been shared) |
| (not available) | run.threadEvaluationTime() | New |
Pass
RunQueryV2ParamsSelect constants (eg. RunQueryV2ParamsSelectName, RunQueryV2ParamsSelectStatus) to Selects to control which fields are populated; unselected fields are zero-valued on the returned struct. Selects defaults to ID only.Before (RunSchema field) | After (Run field) | Notes |
|---|---|---|
run.ID | run.ID | Unchanged |
run.Name | run.Name | Unchanged |
run.RunType | run.RunType | Values changed to uppercase: "LLM", "CHAIN", etc. |
run.Status | run.Status | Values: "SUCCESS", "ERROR", "PENDING" |
run.TraceID | run.TraceID | Unchanged |
run.DottedOrder | run.DottedOrder | Unchanged |
run.AppPath | run.AppPath | Unchanged |
run.StartTime | run.StartTime | Unchanged |
run.EndTime | run.EndTime | Unchanged |
run.Error | run.Error | Unchanged |
run.Events | run.Events | Unchanged; element type is now RunEvent (was map[string]interface{}) |
run.Extra | run.Extra | Unchanged; type is now interface{} (was map[string]interface{}) |
run.FeedbackStats | run.FeedbackStats | Unchanged; element type is now RunFeedbackStat |
run.FirstTokenTime | run.FirstTokenTime | Unchanged |
run.Inputs | run.Inputs | Unchanged; type is now interface{} (was map[string]interface{}) |
run.InputsPreview | run.InputsPreview | Unchanged |
run.Outputs | run.Outputs | Unchanged; type is now interface{} (was map[string]interface{}) |
run.OutputsPreview | run.OutputsPreview | Unchanged |
run.ParentRunIDs | run.ParentRunIDs | Unchanged |
run.PriceModelID | run.PriceModelID | Unchanged |
run.PromptCost | run.PromptCost | Unchanged |
run.PromptCostDetails | run.PromptCostDetails.Raw | Field now wraps the map; access .Raw to get map[string]float64 (was map[string]string) |
run.PromptTokenDetails | run.PromptTokenDetails.Raw | Field now wraps the map; access .Raw to get map[string]int64 (element type unchanged) |
run.PromptTokens | run.PromptTokens | Unchanged |
run.CompletionCost | run.CompletionCost | Unchanged |
run.CompletionCostDetails | run.CompletionCostDetails.Raw | Field now wraps the map; access .Raw to get map[string]float64 (was map[string]string) |
run.CompletionTokenDetails | run.CompletionTokenDetails.Raw | Field now wraps the map; access .Raw to get map[string]int64 (element type unchanged) |
run.CompletionTokens | run.CompletionTokens | Unchanged |
run.TotalCost | run.TotalCost | Unchanged |
run.TotalTokens | run.TotalTokens | Unchanged |
run.ReferenceDatasetID | run.ReferenceDatasetID | Unchanged |
run.ReferenceExampleID | run.ReferenceExampleID | Unchanged |
run.Tags | run.Tags | Unchanged |
run.ThreadID | run.ThreadID | Unchanged |
run.SessionID | run.ProjectID | Renamed |
run.InDataset | run.IsInDataset | Renamed |
run.ChildRunIDs | (removed) | No equivalent |
run.DirectChildRunIDs | (removed) | No equivalent |
run.ExecutionOrder | (removed) | No equivalent |
run.InputsS3URLs | (removed) | Internal storage URL; not exposed in v2 |
run.LastQueuedAt | (removed) | No equivalent |
run.ManifestID | (removed) | Use run.Manifest |
run.ManifestS3ID | (removed) | Internal storage URL; not exposed in v2 |
run.Messages | (removed) | No equivalent |
run.OutputsS3URLs | (removed) | Internal storage URL; not exposed in v2 |
run.ParentRunID | (removed) | Use run.ParentRunIDs |
run.S3URLs | (removed) | Internal storage URL; not exposed in v2 |
run.Serialized | (removed) | Use run.Manifest |
run.ShareToken | (removed) | Use run.ShareURL |
run.TraceFirstReceivedAt | (removed) | No equivalent |
run.TraceMaxStartTime | (removed) | No equivalent |
run.TraceMinStartTime | (removed) | No equivalent |
run.TraceTier | (removed) | No equivalent |
run.TraceUpgrade | (removed) | No equivalent |
run.TtlSeconds | (removed) | No equivalent |
| (not available) | run.Attachments | New: maps attachment filename to pre-signed download URL |
| (not available) | run.ErrorPreview | New: truncated error snippet |
| (not available) | run.IsRoot | New |
| (not available) | run.LatencySeconds | New: wall-clock duration in seconds |
| (not available) | run.Manifest | New: full manifest object (replaces Serialized and ManifestID) |
| (not available) | run.Metadata | New: arbitrary user-defined JSON metadata |
| (not available) | run.ShareURL | New: public share URL (only set when the run has been shared) |
| (not available) | run.ThreadEvaluationTime | New |
Field names in the JSON response match Python
snake_case.Pass SCREAMING_SNAKE_CASE strings in the selects JSON array (eg. "ID", "NAME", "STATUS") to control which fields are populated. Default selects contains only "ID".| Before (v1 response field) | After (v2 response field) | Notes |
|---|---|---|
id | id | Unchanged |
name | name | Unchanged |
run_type | run_type | Values changed to uppercase: "LLM", "CHAIN", etc. |
status | status | Values: "SUCCESS", "ERROR", "PENDING" |
trace_id | trace_id | Unchanged |
dotted_order | dotted_order | Unchanged |
app_path | app_path | Unchanged |
start_time | start_time | Unchanged |
end_time | end_time | Unchanged |
error | error | Unchanged |
events | events | Unchanged |
extra | extra | Unchanged |
feedback_stats | feedback_stats | Unchanged |
first_token_time | first_token_time | Unchanged |
inputs | inputs | Unchanged |
inputs_preview | inputs_preview | Unchanged |
outputs | outputs | Unchanged |
outputs_preview | outputs_preview | Unchanged |
parent_run_ids | parent_run_ids | Unchanged |
price_model_id | price_model_id | Unchanged |
prompt_cost | prompt_cost | Unchanged |
prompt_cost_details | prompt_cost_details.raw | Field now wraps the object; read .raw for the same {category: cost} mapping, now with numeric values (was strings) |
prompt_token_details | prompt_token_details.raw | Field now wraps the object; read .raw for the same {category: count} mapping (values unchanged) |
prompt_tokens | prompt_tokens | Unchanged |
completion_cost | completion_cost | Unchanged |
completion_cost_details | completion_cost_details.raw | Field now wraps the object; read .raw for the same {category: cost} mapping, now with numeric values (was strings) |
completion_token_details | completion_token_details.raw | Field now wraps the object; read .raw for the same {category: count} mapping (values unchanged) |
completion_tokens | completion_tokens | Unchanged |
total_cost | total_cost | Unchanged |
total_tokens | total_tokens | Unchanged |
reference_dataset_id | reference_dataset_id | Unchanged |
reference_example_id | reference_example_id | Unchanged |
tags | tags | Unchanged |
thread_id | thread_id | Unchanged |
session_id | project_id | Renamed |
in_dataset | is_in_dataset | Renamed |
child_run_ids | (removed) | No equivalent |
direct_child_run_ids | (removed) | No equivalent |
execution_order | (removed) | No equivalent |
inputs_s3_urls | (removed) | Internal storage URL; not exposed in v2 |
last_queued_at | (removed) | No equivalent |
manifest_id | (removed) | Use manifest |
manifest_s3_id | (removed) | Internal storage URL; not exposed in v2 |
messages | (removed) | No equivalent |
outputs_s3_urls | (removed) | Internal storage URL; not exposed in v2 |
parent_run_id | (removed) | Use parent_run_ids |
s3_urls | (removed) | Internal storage URL; not exposed in v2 |
serialized | (removed) | Use manifest |
share_token | (removed) | Use share_url |
trace_first_received_at | (removed) | No equivalent |
trace_max_start_time | (removed) | No equivalent |
trace_min_start_time | (removed) | No equivalent |
trace_tier | (removed) | No equivalent |
trace_upgrade | (removed) | No equivalent |
ttl_seconds | (removed) | No equivalent |
| (not available) | attachments | New: maps attachment filename to pre-signed download URL |
| (not available) | error_preview | New: truncated error snippet |
| (not available) | is_root | New |
| (not available) | latency_seconds | New: wall-clock duration in seconds |
| (not available) | manifest | New: full manifest object (replaces serialized and manifest_id) |
| (not available) | metadata | New: previously nested under extra.metadata |
| (not available) | share_url | New: public share URL (only set when the run has been shared) |
| (not available) | thread_evaluation_time | New |
Examples
List all runs in a project
- Python
- TypeScript
- Java
- Go
- cURL
runs.query does not accept a project name directly. Resolve the project UUID with client.aread_project() first, then pass it as a string in project_ids.- Before
- After
Before
from langsmith import Client
client = Client()
runs = client.list_runs(project_name="default")
After
import asyncio
from langsmith import Client
async def main():
client = Client()
project = await client.aread_project(project_name="default")
runs = client.runs.query(project_ids=[str(project.id)])
asyncio.run(main())
client.runs.query does not accept a project name directly. Resolve the project UUID with client.readProject() first, then pass it as a string in project_ids.- Before
- After
Before
import { Client } from "langsmith";
const client = new Client();
const runs = client.listRuns({ projectName: "default" });
After
import { Client } from "langsmith";
const client = new Client();
const project = await client.readProject({ projectName: "default" });
const runs = client.runs.query({ project_ids: [project.id] });
queryV2() does not accept a project name directly. Resolve the project UUID with client.sessions().list() first, then pass it as a string in projectIds().- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryParams
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val runs = client.runs().query(
RunQueryParams.builder().addSession(project.id()).build()
).items()
After
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryV2Params
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val runs = client.runs().queryV2(
RunQueryV2Params.builder().addProjectId(project.id()).build()
).items()
QueryV2() does not accept a project name directly. Resolve the project UUID with client.Sessions.List() first, then pass it as a string in ProjectIDs.- Before
- After
Before
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runs, err := client.Runs.Query(ctx, langsmith.RunQueryParams{
Session: langsmith.F([]string{project.ID}),
})
After
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runs, err := client.Runs.QueryV2(ctx, langsmith.RunQueryV2Params{
ProjectIDs: langsmith.F([]string{project.ID}),
})
POST /v2/runs/query does not accept a project name directly. Resolve the project UUID with a GET /api/v1/sessions request first, then pass it as a string in project_ids.- Before
- After
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
curl -X POST "https://api.smith.langchain.com/api/v1/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" '{"session": [$pid]}')"
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
curl -X POST "https://api.smith.langchain.com/v2/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" '{"project_ids": [$pid]}')"
Selecting fields
- Python
- TypeScript
- Java
- Go
- cURL
list_runs returns a default set of fields with no selection needed. runs.query returns only id by default—pass selects=[...] to request more. Field names are now uppercase ("name" → "NAME").- Before
- After
Before
from langsmith import Client
client = Client()
# returns a default set of fields; no explicit selection needed
runs = client.list_runs(project_name="default")
for run in runs:
print(run.id, run.name, run.run_type, run.status, run.start_time, run.inputs, run.error)
After
import asyncio
from langsmith import Client
async def main():
client = Client()
project = await client.aread_project(project_name="default")
# must explicitly list every field needed; default returns only id
async for run in client.runs.query(
project_ids=[str(project.id)],
selects=["ID", "NAME", "RUN_TYPE", "STATUS", "START_TIME", "INPUTS", "ERROR"],
):
print(run.id, run.name, run.run_type, run.status, run.start_time, run.inputs, run.error)
asyncio.run(main())
listRuns returns a default set of fields with no selection needed. client.runs.query returns only id by default—pass selects: [...] to request more. Field names are now uppercase ("name" → "NAME").- Before
- After
Before
import { Client } from "langsmith";
const client = new Client();
// returns a default set of fields; no explicit selection needed
const runs = client.listRuns({ projectName: "default" });
for await (const run of runs) {
console.log(run.id, run.name, run.run_type, run.status, run.start_time, run.inputs, run.error);
}
After
import { Client } from "langsmith";
const client = new Client();
const project = await client.readProject({ projectName: "default" });
// must explicitly list every field needed; default returns only id
for await (const run of client.runs.query({
project_ids: [project.id],
selects: ["ID", "NAME", "RUN_TYPE", "STATUS", "START_TIME", "INPUTS", "ERROR"],
})) {
console.log(run.id, run.name, run.run_type, run.status, run.start_time, run.inputs, run.error);
}
query() returns a default set of fields with no selection needed. queryV2() returns only id by default—call .addSelect(RunQueryV2Params.Select.X) for each field you need.- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryParams
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
// returns a default set of fields; no explicit selection needed
val runs = client.runs().query(
RunQueryParams.builder().addSession(project.id()).build()
).items()
for (run in runs) {
println("${run.id()} ${run.name()} ${run.runType()} ${run.status()} ${run.startTime()} ${run.inputs()} ${run.error()}")
}
After
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryV2Params
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
// must explicitly list every field needed; default returns only id
val runs = client.runs().queryV2(
RunQueryV2Params.builder()
.addProjectId(project.id())
.addSelect(RunQueryV2Params.Select.ID)
.addSelect(RunQueryV2Params.Select.NAME)
.addSelect(RunQueryV2Params.Select.RUN_TYPE)
.addSelect(RunQueryV2Params.Select.STATUS)
.addSelect(RunQueryV2Params.Select.START_TIME)
.addSelect(RunQueryV2Params.Select.INPUTS)
.addSelect(RunQueryV2Params.Select.ERROR)
.build()
).items()
for (run in runs) {
println("${run.id()} ${run.name()} ${run.runType()} ${run.status()} ${run.startTime()} ${run.inputs()} ${run.error()}")
}
Query returns a default set of fields with no selection needed. QueryV2 returns only ID by default—pass Selects with the uppercase field constants you need (e.g. RunQueryV2ParamsSelectName).- Before
- After
Before
package main
import (
"context"
"fmt"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
// returns a default set of fields; no explicit selection needed
runs, err := client.Runs.Query(ctx, langsmith.RunQueryParams{
Session: langsmith.F([]string{project.ID}),
})
for _, run := range runs.Runs {
fmt.Println(run.ID, run.Name, run.RunType, run.Status, run.StartTime, run.Inputs, run.Error)
}
After
package main
import (
"context"
"fmt"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
// must explicitly list every field needed; default returns only id
runs, err := client.Runs.QueryV2(ctx, langsmith.RunQueryV2Params{
ProjectIDs: langsmith.F([]string{project.ID}),
Selects: langsmith.F([]langsmith.RunQueryV2ParamsSelect{
langsmith.RunQueryV2ParamsSelectID,
langsmith.RunQueryV2ParamsSelectName,
langsmith.RunQueryV2ParamsSelectRunType,
langsmith.RunQueryV2ParamsSelectStatus,
langsmith.RunQueryV2ParamsSelectStartTime,
langsmith.RunQueryV2ParamsSelectInputs,
langsmith.RunQueryV2ParamsSelectError,
}),
})
for _, run := range runs.Items {
fmt.Println(run.ID, run.Name, run.RunType, run.Status, run.StartTime, run.Inputs, run.Error)
}
POST /api/v1/runs/query returns a default set of fields with no selection needed. POST /v2/runs/query returns only id by default—pass selects with the uppercase field names you need (e.g. "NAME").- Before
- After
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
curl -X POST "https://api.smith.langchain.com/api/v1/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" '{"session": [$pid]}')"
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
curl -X POST "https://api.smith.langchain.com/v2/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" '{"project_ids": [$pid], "selects": ["ID", "NAME", "RUN_TYPE", "STATUS", "START_TIME", "INPUTS", "ERROR"]}')"
Filter by run type and time range
- Python
- TypeScript
- Java
- Go
- cURL
start_time is renamed to min_start_time, and run_type values are now uppercase ("llm" → "LLM").- Before
- After
Before
from datetime import datetime, timedelta
from langsmith import Client
client = Client()
runs = client.list_runs(
project_name="default",
start_time=datetime.now() - timedelta(days=1),
run_type="llm",
)
After
import asyncio
from datetime import datetime, timedelta
from langsmith import Client
async def main():
client = Client()
project = await client.aread_project(project_name="default")
runs = client.runs.query(
project_ids=[str(project.id)],
min_start_time=datetime.now() - timedelta(days=1),
run_type="LLM",
)
asyncio.run(main())
startTime (camelCase) becomes min_start_time (snake_case, matching the v2 request body), and runType values are now uppercase ("llm" → "LLM").- Before
- After
Before
import { Client } from "langsmith";
const client = new Client();
const runs = client.listRuns({
projectName: "default",
startTime: new Date(Date.now() - 24 * 60 * 60 * 1000),
runType: "llm",
});
After
import { Client } from "langsmith";
const client = new Client();
const project = await client.readProject({ projectName: "default" });
const oneDayAgo = new Date(Date.now() - 24 * 60 * 60 * 1000);
const runs = client.runs.query({
project_ids: [project.id],
min_start_time: oneDayAgo.toISOString(),
run_type: "LLM",
});
.startTime() is renamed to .minStartTime(), and .runType() now takes the new RunQueryV2Params.RunType enum instead of RunTypeEnum.- Before
- After
Before
import java.time.OffsetDateTime
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryParams
import com.langchain.smith.models.runs.RunTypeEnum
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val runs = client.runs().query(
RunQueryParams.builder()
.addSession(project.id())
.startTime(OffsetDateTime.now().minusDays(1))
.runType(RunTypeEnum.LLM)
.build()
).items()
After
import java.time.OffsetDateTime
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryV2Params
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val runs = client.runs().queryV2(
RunQueryV2Params.builder()
.addProjectId(project.id())
.minStartTime(OffsetDateTime.now().minusDays(1))
.runType(RunQueryV2Params.RunType.LLM)
.build()
).items()
StartTime is renamed to MinStartTime, and RunType now takes the new RunQueryV2ParamsRunType enum instead of RunTypeEnum.- Before
- After
Before
package main
import (
"context"
"time"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runs, err := client.Runs.Query(ctx, langsmith.RunQueryParams{
Session: langsmith.F([]string{project.ID}),
StartTime: langsmith.F(time.Now().Add(-24 * time.Hour)),
RunType: langsmith.F(langsmith.RunTypeEnumLlm),
})
After
package main
import (
"context"
"time"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runs, err := client.Runs.QueryV2(ctx, langsmith.RunQueryV2Params{
ProjectIDs: langsmith.F([]string{project.ID}),
MinStartTime: langsmith.F(time.Now().Add(-24 * time.Hour)),
RunType: langsmith.F(langsmith.RunQueryV2ParamsRunTypeLlm),
})
start_time is renamed to min_start_time, and run_type values are now uppercase ("llm" → "LLM").- Before
- After
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
curl -X POST "https://api.smith.langchain.com/api/v1/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" '{"session": [$pid], "run_type": "llm", "start_time": "2025-01-01T00:00:00Z"}')"
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
curl -X POST "https://api.smith.langchain.com/v2/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" '{"project_ids": [$pid], "run_type": "LLM", "min_start_time": "2025-01-01T00:00:00Z"}')"
Filter root runs only
- Python
- TypeScript
- Java
- Go
- cURL
is_root is unchanged.- Before
- After
Before
from langsmith import Client
client = Client()
runs = client.list_runs(project_name="default", is_root=True)
After
import asyncio
from langsmith import Client
async def main():
client = Client()
project = await client.aread_project(project_name="default")
runs = client.runs.query(project_ids=[str(project.id)], is_root=True)
asyncio.run(main())
isRoot (camelCase) becomes is_root (snake_case, matching the v2 request body).- Before
- After
Before
import { Client } from "langsmith";
const client = new Client();
const runs = client.listRuns({ projectName: "default", isRoot: true });
After
import { Client } from "langsmith";
const client = new Client();
const project = await client.readProject({ projectName: "default" });
const runs = client.runs.query({
project_ids: [project.id],
is_root: true,
});
.isRoot() is unchanged.- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryParams
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val runs = client.runs().query(
RunQueryParams.builder().addSession(project.id()).isRoot(true).build()
).items()
After
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryV2Params
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val runs = client.runs().queryV2(
RunQueryV2Params.builder().addProjectId(project.id()).isRoot(true).build()
).items()
IsRoot is unchanged.- Before
- After
Before
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runs, err := client.Runs.Query(ctx, langsmith.RunQueryParams{
Session: langsmith.F([]string{project.ID}),
IsRoot: langsmith.F(true),
})
After
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runs, err := client.Runs.QueryV2(ctx, langsmith.RunQueryV2Params{
ProjectIDs: langsmith.F([]string{project.ID}),
IsRoot: langsmith.F(true),
})
is_root is unchanged.- Before
- After
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
curl -X POST "https://api.smith.langchain.com/api/v1/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" '{"session": [$pid], "is_root": true}')"
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
curl -X POST "https://api.smith.langchain.com/v2/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" '{"project_ids": [$pid], "is_root": true}')"
Fetch runs by ID list
- Python
- TypeScript
- Java
- Go
- cURL
id=[...] is renamed to ids=[...]. project_ids is now required even when filtering by run IDs—v1 allowed omitting the project context.- Before
- After
Before
from langsmith import Client
client = Client()
runs = client.list_runs(id=["<run-id-1>", "<run-id-2>"])
After
import asyncio
from langsmith import Client
async def main():
client = Client()
project = await client.aread_project(project_name="default")
runs = client.runs.query(
project_ids=[str(project.id)],
ids=["<run-id-1>", "<run-id-2>"],
)
asyncio.run(main())
id: [...] is renamed to ids: [...]. project_ids is now required even when filtering by run IDs—v1 allowed omitting the project context.- Before
- After
Before
import { Client } from "langsmith";
const client = new Client();
const runs = client.listRuns({ id: ["<run-id-1>", "<run-id-2>"] });
After
import { Client } from "langsmith";
const client = new Client();
const project = await client.readProject({ projectName: "default" });
const runs = client.runs.query({
project_ids: [project.id],
ids: ["<run-id-1>", "<run-id-2>"],
});
.addId(...) is unchanged—call it once per run ID. .addProjectId(...) is now required even when filtering by run IDs—v1 allowed omitting the project context.- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryParams
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
var runId1 = "<run-id-1>"
var runId2 = "<run-id-2>"
val runs = client.runs().query(
RunQueryParams.builder()
.addSession(project.id())
.addId(runId1)
.addId(runId2)
.build()
).items()
After
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryV2Params
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val runs = client.runs().queryV2(
RunQueryV2Params.builder()
.addProjectId(project.id())
.addId("<run-id-1>")
.addId("<run-id-2>")
.build()
).items()
ID: [...] is renamed to IDs: [...]. ProjectIDs is now required even when filtering by run IDs—v1 allowed omitting the project context.- Before
- After
Before
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runID1 := "<run-id-1>"
runID2 := "<run-id-2>"
runs, err := client.Runs.Query(ctx, langsmith.RunQueryParams{
Session: langsmith.F([]string{project.ID}),
ID: langsmith.F([]string{runID1, runID2}),
})
After
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runID1 := "<run-id-1>"
runID2 := "<run-id-2>"
runs, err := client.Runs.QueryV2(ctx, langsmith.RunQueryV2Params{
ProjectIDs: langsmith.F([]string{project.ID}),
IDs: langsmith.F([]string{runID1, runID2}),
})
id is renamed to ids. project_ids is now required in the request body even when filtering by run IDs—v1 allowed omitting the project context.- Before
- After
RUN_ID_1="<run-id-1>"
RUN_ID_2="<run-id-2>"
curl -X POST "https://api.smith.langchain.com/api/v1/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg r1 "$RUN_ID_1" --arg r2 "$RUN_ID_2" '{"id": [$r1, $r2]}')"
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
RUN_ID_1="<run-id-1>"
RUN_ID_2="<run-id-2>"
curl -X POST "https://api.smith.langchain.com/v2/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" --arg r1 "$RUN_ID_1" --arg r2 "$RUN_ID_2" '{"project_ids": [$pid], "ids": [$r1, $r2]}')"
Iterate through runs
- Python
- TypeScript
- Java
- Go
- cURL
list_runs auto-paginates transparently, fetching up to 100 runs per API call and stopping once limit results are returned. runs.query does not accept a total limit; iterate with async for and break once you have enough, or use the returned page’s has_next_page()/get_next_page() for manual page-by-page control.- Before
- After
Before
from langsmith import Client
client = Client()
runs = client.list_runs(project_name="default", limit=150)
After
import asyncio
from langsmith import Client
async def main():
client = Client()
project = await client.aread_project(project_name="default")
runs = []
async for run in client.runs.query(
project_ids=[str(project.id)],
):
runs.append(run)
if len(runs) >= 150:
break
asyncio.run(main())
listRuns auto-paginates transparently. client.runs.query returns an async iterable of individual runs—use for await and break once you have enough.- Before
- After
Before
import { Client } from "langsmith";
const client = new Client();
const runs: unknown[] = [];
for await (const run of client.listRuns({ projectName: "default" })) {
runs.push(run);
if (runs.length >= 150) break;
}
After
import { Client } from "langsmith";
const client = new Client();
const project = await client.readProject({ projectName: "default" });
const runs: unknown[] = [];
for await (const run of client.runs.query({
project_ids: [project.id],
})) {
runs.push(run);
if (runs.length >= 150) break;
}
.autoPager() is used the same way on both query() and queryV2()—break out of the loop once you have enough runs.- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryParams
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val runs = mutableListOf<Any>()
for (run in client.runs().query(
RunQueryParams.builder().addSession(project.id()).build()
).autoPager()) {
runs.add(run)
if (runs.size >= 150) break
}
After
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryV2Params
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val runs = mutableListOf<Any>()
for (run in client.runs().queryV2(
RunQueryV2Params.builder().addProjectId(project.id()).build()
).autoPager()) {
runs.add(run)
if (runs.size >= 150) break
}
QueryAutoPaging is renamed to QueryV2AutoPaging; both use the same iter.Next()/iter.Current() pattern—break out of the loop once you have enough runs.- Before
- After
Before
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runs := []langsmith.RunSchema{}
iter := client.Runs.QueryAutoPaging(ctx, langsmith.RunQueryParams{
Session: langsmith.F([]string{project.ID}),
})
for iter.Next() {
runs = append(runs, iter.Current())
if len(runs) >= 150 {
break
}
}
After
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runs := []langsmith.Run{}
iter := client.Runs.QueryV2AutoPaging(ctx, langsmith.RunQueryV2Params{
ProjectIDs: langsmith.F([]string{project.ID}),
})
for iter.Next() {
runs = append(runs, iter.Current())
if len(runs) >= 150 {
break
}
}
The v1 API returns all matching runs in one response with no cursor. The v2 API paginates—pass the
cursor from a response’s next_cursor field to fetch the next page.- Before
- After
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
curl -X POST "https://api.smith.langchain.com/api/v1/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" '{"session": [$pid], "limit": 150}')"
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
# Fetch pages, passing the cursor from each response's next_cursor field
# to fetch the next page, until 150 runs are collected or pages run out.
TOTAL=0
CURSOR=""
while :; do
BODY=$(jq -n --arg pid "$PROJECT_ID" --arg cursor "$CURSOR" \
'if $cursor == "" then {"project_ids": [$pid]} else {"project_ids": [$pid], "cursor": $cursor} end')
RESPONSE=$(curl -s -X POST "https://api.smith.langchain.com/v2/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$BODY")
TOTAL=$((TOTAL + $(echo "$RESPONSE" | jq '.items | length')))
CURSOR=$(echo "$RESPONSE" | jq -r '.next_cursor // empty')
[ "$TOTAL" -lt 150 ] && [ -n "$CURSOR" ] || break
done
Filter runs with errors
- Python
- TypeScript
- Java
- Go
- cURL
error=True/False is renamed to has_error=True/False.- Before
- After
Before
from langsmith import Client
client = Client()
runs = client.list_runs(project_name="default", error=True)
After
import asyncio
from langsmith import Client
async def main():
client = Client()
project = await client.aread_project(project_name="default")
runs = client.runs.query(project_ids=[str(project.id)], has_error=True)
asyncio.run(main())
error: true/false is renamed to has_error: true/false.- Before
- After
Before
import { Client } from "langsmith";
const client = new Client();
const runs = client.listRuns({ projectName: "default", error: true });
After
import { Client } from "langsmith";
const client = new Client();
const project = await client.readProject({ projectName: "default" });
const runs = client.runs.query({
project_ids: [project.id],
has_error: true,
});
.error(true/false) is renamed to .hasError(true/false).- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryParams
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val runs = client.runs().query(
RunQueryParams.builder().addSession(project.id()).error(true).build()
).items()
After
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryV2Params
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val runs = client.runs().queryV2(
RunQueryV2Params.builder().addProjectId(project.id()).hasError(true).build()
).items()
Error is renamed to HasError.- Before
- After
Before
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runs, err := client.Runs.Query(ctx, langsmith.RunQueryParams{
Session: langsmith.F([]string{project.ID}),
Error: langsmith.F(true),
})
After
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runs, err := client.Runs.QueryV2(ctx, langsmith.RunQueryV2Params{
ProjectIDs: langsmith.F([]string{project.ID}),
HasError: langsmith.F(true),
})
error is renamed to has_error.- Before
- After
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
curl -X POST "https://api.smith.langchain.com/api/v1/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" '{"session": [$pid], "error": true}')"
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
curl -X POST "https://api.smith.langchain.com/v2/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" '{"project_ids": [$pid], "has_error": true}')"
Filter by metadata
- Python
- TypeScript
- Java
- Go
- cURL
The
filter string syntax is unchanged: eq(metadata_key, ...) checks for key presence, combined with eq(metadata_value, ...) to match a specific value.- Before
- After
Before
from langsmith import Client
client = Client()
filter_str = 'and(eq(metadata_key, "user_id"), eq(metadata_value, "u_123"))'
runs = client.list_runs(project_name="default", filter=filter_str)
After
import asyncio
from langsmith import Client
async def main():
client = Client()
filter_str = 'and(eq(metadata_key, "user_id"), eq(metadata_value, "u_123"))'
project = await client.aread_project(project_name="default")
runs = client.runs.query(project_ids=[str(project.id)], filter=filter_str)
asyncio.run(main())
The
filter string syntax is unchanged: eq(metadata_key, ...) checks for key presence, combined with eq(metadata_value, ...) to match a specific value.- Before
- After
Before
import { Client } from "langsmith";
const client = new Client();
const filterStr = 'and(eq(metadata_key, "user_id"), eq(metadata_value, "u_123"))';
const runs = client.listRuns({ projectName: "default", filter: filterStr });
After
import { Client } from "langsmith";
const client = new Client();
const filterStr = 'and(eq(metadata_key, "user_id"), eq(metadata_value, "u_123"))';
const project = await client.readProject({ projectName: "default" });
const runs = client.runs.query({
project_ids: [project.id],
filter: filterStr,
});
The
.filter(...) string syntax is unchanged: eq(metadata_key, ...) checks for key presence, combined with eq(metadata_value, ...) to match a specific value.- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryParams
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val filterStr = "and(eq(metadata_key, \"user_id\"), eq(metadata_value, \"u_123\"))"
val runs = client.runs().query(
RunQueryParams.builder().addSession(project.id()).filter(filterStr).build()
).items()
After
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryV2Params
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val filterStr = "and(eq(metadata_key, \"user_id\"), eq(metadata_value, \"u_123\"))"
val runs = client.runs().queryV2(
RunQueryV2Params.builder().addProjectId(project.id()).filter(filterStr).build()
).items()
The
Filter string syntax is unchanged: eq(metadata_key, ...) checks for key presence, combined with eq(metadata_value, ...) to match a specific value.- Before
- After
Before
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
filterStr := `and(eq(metadata_key, "user_id"), eq(metadata_value, "u_123"))`
runs, err := client.Runs.Query(ctx, langsmith.RunQueryParams{
Session: langsmith.F([]string{project.ID}),
Filter: langsmith.F(filterStr),
})
After
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
filterStr := `and(eq(metadata_key, "user_id"), eq(metadata_value, "u_123"))`
runs, err := client.Runs.QueryV2(ctx, langsmith.RunQueryV2Params{
ProjectIDs: langsmith.F([]string{project.ID}),
Filter: langsmith.F(filterStr),
})
The
filter string syntax is unchanged: eq(metadata_key, ...) checks for key presence, combined with eq(metadata_value, ...) to match a specific value.- Before
- After
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
FILTER='and(eq(metadata_key, "user_id"), eq(metadata_value, "u_123"))'
curl -X POST "https://api.smith.langchain.com/api/v1/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" --arg f "$FILTER" '{"session": [$pid], "filter": $f}')"
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
FILTER='and(eq(metadata_key, "user_id"), eq(metadata_value, "u_123"))'
curl -X POST "https://api.smith.langchain.com/v2/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" --arg f "$FILTER" '{"project_ids": [$pid], "filter": $f}')"
Complex boolean filters
- Python
- TypeScript
- Java
- Go
- cURL
Nested
and() / or() filter expressions are unchanged.- Before
- After
Before
from langsmith import Client
client = Client()
filter_str = (
'and(gt(start_time, "2023-07-15T12:34:56Z"),'
' or(neq(status, "error"),'
' and(eq(feedback_key, "Correctness"), eq(feedback_score, 0.0))))'
)
runs = client.list_runs(project_name="default", filter=filter_str)
After
import asyncio
from langsmith import Client
async def main():
client = Client()
filter_str = (
'and(gt(start_time, "2023-07-15T12:34:56Z"),'
' or(neq(status, "error"),'
' and(eq(feedback_key, "Correctness"), eq(feedback_score, 0.0))))'
)
project = await client.aread_project(project_name="default")
runs = client.runs.query(project_ids=[str(project.id)], filter=filter_str)
asyncio.run(main())
Nested
and() / or() filter expressions are unchanged.- Before
- After
Before
import { Client } from "langsmith";
const client = new Client();
const filterStr =
'and(gt(start_time, "2023-07-15T12:34:56Z"),' +
' or(neq(status, "error"),' +
' and(eq(feedback_key, "Correctness"), eq(feedback_score, 0.0))))';
const runs = client.listRuns({ projectName: "default", filter: filterStr });
After
import { Client } from "langsmith";
const client = new Client();
const filterStr =
'and(gt(start_time, "2023-07-15T12:34:56Z"),' +
' or(neq(status, "error"),' +
' and(eq(feedback_key, "Correctness"), eq(feedback_score, 0.0))))';
const project = await client.readProject({ projectName: "default" });
const runs = client.runs.query({
project_ids: [project.id],
filter: filterStr,
});
Nested
and() / or() filter expressions are unchanged.- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryParams
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val filterStr = "and(gt(start_time, \"2023-07-15T12:34:56Z\")," +
" or(neq(status, \"error\")," +
" and(eq(feedback_key, \"Correctness\"), eq(feedback_score, 0.0))))"
val runs = client.runs().query(
RunQueryParams.builder().addSession(project.id()).filter(filterStr).build()
).items()
After
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryV2Params
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val filterStr = "and(gt(start_time, \"2023-07-15T12:34:56Z\")," +
" or(neq(status, \"error\")," +
" and(eq(feedback_key, \"Correctness\"), eq(feedback_score, 0.0))))"
val runs = client.runs().queryV2(
RunQueryV2Params.builder().addProjectId(project.id()).filter(filterStr).build()
).items()
Nested
and() / or() filter expressions are unchanged.- Before
- After
Before
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
filterStr := `and(gt(start_time, "2023-07-15T12:34:56Z"), or(neq(status, "error"), and(eq(feedback_key, "Correctness"), eq(feedback_score, 0.0))))`
runs, err := client.Runs.Query(ctx, langsmith.RunQueryParams{
Session: langsmith.F([]string{project.ID}),
Filter: langsmith.F(filterStr),
})
After
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
filterStr := `and(gt(start_time, "2023-07-15T12:34:56Z"), or(neq(status, "error"), and(eq(feedback_key, "Correctness"), eq(feedback_score, 0.0))))`
runs, err := client.Runs.QueryV2(ctx, langsmith.RunQueryV2Params{
ProjectIDs: langsmith.F([]string{project.ID}),
Filter: langsmith.F(filterStr),
})
Nested
and() / or() filter expressions are unchanged.- Before
- After
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
FILTER='and(gt(start_time, "2023-07-15T12:34:56Z"), or(neq(status, "error"), and(eq(feedback_key, "Correctness"), eq(feedback_score, 0.0))))'
curl -X POST "https://api.smith.langchain.com/api/v1/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" --arg f "$FILTER" '{"session": [$pid], "filter": $f}')"
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
FILTER='and(gt(start_time, "2023-07-15T12:34:56Z"), or(neq(status, "error"), and(eq(feedback_key, "Correctness"), eq(feedback_score, 0.0))))'
curl -X POST "https://api.smith.langchain.com/v2/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg pid "$PROJECT_ID" --arg f "$FILTER" '{"project_ids": [$pid], "filter": $f}')"
Scoped filters: filter, trace_filter, tree_filter
- Python
- TypeScript
- Java
- Go
- cURL
filter, trace_filter, and tree_filter are unchanged. filter applies to the matched run, trace_filter to the root of its trace, and tree_filter to other runs in the trace tree (siblings and children).- Before
- After
Before
from langsmith import Client
client = Client()
runs = client.list_runs(
project_name="default",
filter='eq(name, "RetrieveDocs")',
trace_filter='and(eq(feedback_key, "user_score"), eq(feedback_score, 1))',
tree_filter='eq(name, "ExpandQuery")',
)
After
import asyncio
from langsmith import Client
async def main():
client = Client()
project = await client.aread_project(project_name="default")
runs = client.runs.query(
project_ids=[str(project.id)],
filter='eq(name, "RetrieveDocs")',
trace_filter='and(eq(feedback_key, "user_score"), eq(feedback_score, 1))',
tree_filter='eq(name, "ExpandQuery")',
)
asyncio.run(main())
filter, trace_filter, and tree_filter are unchanged. filter applies to the matched run, trace_filter to the root of its trace, and tree_filter to other runs in the trace tree (siblings and children).- Before
- After
Before
import { Client } from "langsmith";
const client = new Client();
const runs = client.listRuns({
projectName: "default",
filter: 'eq(name, "RetrieveDocs")',
traceFilter: 'and(eq(feedback_key, "user_score"), eq(feedback_score, 1))',
treeFilter: 'eq(name, "ExpandQuery")',
});
After
import { Client } from "langsmith";
const client = new Client();
const project = await client.readProject({ projectName: "default" });
const runs = client.runs.query({
project_ids: [project.id],
filter: 'eq(name, "RetrieveDocs")',
trace_filter: 'and(eq(feedback_key, "user_score"), eq(feedback_score, 1))',
tree_filter: 'eq(name, "ExpandQuery")',
});
.filter(), .traceFilter(), and .treeFilter() are unchanged. filter applies to the matched run, traceFilter to the root of its trace, and treeFilter to other runs in the trace tree (siblings and children).- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryParams
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val runs = client.runs().query(
RunQueryParams.builder()
.addSession(project.id())
.filter("eq(name, \"RetrieveDocs\")")
.traceFilter("and(eq(feedback_key, \"user_score\"), eq(feedback_score, 1))")
.treeFilter("eq(name, \"ExpandQuery\")")
.build()
).items()
After
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunQueryV2Params
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
val runs = client.runs().queryV2(
RunQueryV2Params.builder()
.addProjectId(project.id())
.filter("eq(name, \"RetrieveDocs\")")
.traceFilter("and(eq(feedback_key, \"user_score\"), eq(feedback_score, 1))")
.treeFilter("eq(name, \"ExpandQuery\")")
.build()
).items()
Filter, TraceFilter, and TreeFilter are unchanged. Filter applies to the matched run, TraceFilter to the root of its trace, and TreeFilter to other runs in the trace tree (siblings and children).- Before
- After
Before
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runs, err := client.Runs.Query(ctx, langsmith.RunQueryParams{
Session: langsmith.F([]string{project.ID}),
Filter: langsmith.F(`eq(name, "RetrieveDocs")`),
TraceFilter: langsmith.F(`and(eq(feedback_key, "user_score"), eq(feedback_score, 1))`),
TreeFilter: langsmith.F(`eq(name, "ExpandQuery")`),
})
After
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
sessions, err := client.Sessions.List(ctx, langsmith.SessionListParams{
Name: langsmith.F("default"),
Limit: langsmith.F(int64(1)),
})
project := sessions.Items[0]
runs, err := client.Runs.QueryV2(ctx, langsmith.RunQueryV2Params{
ProjectIDs: langsmith.F([]string{project.ID}),
Filter: langsmith.F(`eq(name, "RetrieveDocs")`),
TraceFilter: langsmith.F(`and(eq(feedback_key, "user_score"), eq(feedback_score, 1))`),
TreeFilter: langsmith.F(`eq(name, "ExpandQuery")`),
})
filter, trace_filter, and tree_filter are unchanged. filter applies to the matched run, trace_filter to the root of its trace, and tree_filter to other runs in the trace tree (siblings and children).- Before
- After
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
FILTER='eq(name, "RetrieveDocs")'
TRACE_FILTER='and(eq(feedback_key, "user_score"), eq(feedback_score, 1))'
TREE_FILTER='eq(name, "ExpandQuery")'
curl -X POST "https://api.smith.langchain.com/api/v1/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n \
--arg pid "$PROJECT_ID" \
--arg f "$FILTER" \
--arg tf "$TRACE_FILTER" \
--arg treef "$TREE_FILTER" \
'{"session": [$pid], "filter": $f, "trace_filter": $tf, "tree_filter": $treef}')"
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
FILTER='eq(name, "RetrieveDocs")'
TRACE_FILTER='and(eq(feedback_key, "user_score"), eq(feedback_score, 1))'
TREE_FILTER='eq(name, "ExpandQuery")'
curl -X POST "https://api.smith.langchain.com/v2/runs/query" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n \
--arg pid "$PROJECT_ID" \
--arg f "$FILTER" \
--arg tf "$TRACE_FILTER" \
--arg treef "$TREE_FILTER" \
'{"project_ids": [$pid], "filter": $f, "trace_filter": $tf, "tree_filter": $treef}')"
Runs: retrieve
Fetch a single run by ID. Returns only the run ID by default—specify a field selection list to retrieve additional data.Main changes
Method name
- Python
- TypeScript
- Java
- Go
- cURL
| Before | After |
|---|---|
client.read_run() | client.runs.retrieve() |
client.runs.retrieve() is now async. Call it with await.| Before | After |
|---|---|
client.readRun() | client.runs.retrieve() |
| Before | After |
|---|---|
client.runs().retrieve() | client.runs().retrieveV2() |
| Before | After |
|---|---|
client.Runs.Get() | client.Runs.GetV2() |
| Before | After |
|---|---|
GET /api/v1/runs/{run_id} | GET /v2/runs/{run_id} |
Query parameters
- Python
- TypeScript
- Java
- Go
- cURL
runs.retrieve requires two new fields—project_id and start_time—that read_run did not need. Both are required to locate the run in SmithDB.Before (read_run) | After (runs.retrieve) | Notes |
|---|---|---|
run_id | run_id | Unchanged |
load_child_runs | (removed) | No equivalent |
| (not available) | project_id | Required—UUID of the project that owns the run |
| (not available) | start_time | Required—run’s start time (RFC3339), used with project_id to locate the run |
| (all fields returned by default) | selects | Field projection; defaults to ["ID"] only; field names are uppercase |
client.runs.retrieve requires two new fields—project_id and start_time—that readRun did not need. Both are required to locate the run in SmithDB.Before (readRun) | After (client.runs.retrieve) | Notes |
|---|---|---|
runId | runId | Unchanged (positional parameter) |
options.loadChildRuns | (removed) | No equivalent |
| (not available) | project_id | Required—snake_case; UUID of the project that owns the run |
| (not available) | start_time | Required—snake_case; run’s start time (RFC3339), used to locate the run |
| (all fields returned by default) | selects | Field projection; defaults to ["ID"] only; field names are uppercase |
retrieveV2() requires projectId(), which replaces the removed sessionId(), and makes startTime() required (previously optional). Both are needed to locate the run in SmithDB.Before (RunRetrieveParams) | After (RunRetrieveV2Params) | Notes |
|---|---|---|
runId() | runId() | Unchanged |
sessionId() | (removed) | Replaced by projectId() |
startTime() | startTime() | Now required; used with projectId() to locate the run in SmithDB |
excludeS3StoredAttributes() | (removed) | No equivalent |
excludeSerialized() | (removed) | No equivalent |
includeMessages() | (removed) | No equivalent |
| (not available) | projectId() | Required—UUID of the project that owns the run |
| (all fields returned by default) | selects() | Field projection; defaults to ["ID"] only; field names are uppercase |
GetV2() requires ProjectID, which replaces the removed SessionID, and makes StartTime required (previously optional). Both are needed to locate the run in SmithDB.Before (RunGetParams) | After (RunGetV2Params) | Notes |
|---|---|---|
runID (positional) | runID (positional) | Unchanged |
ExcludeS3StoredAttributes | (removed) | No equivalent |
ExcludeSerialized | (removed) | No equivalent |
IncludeMessages | (removed) | No equivalent |
SessionID | (removed) | Replaced by ProjectID |
StartTime | StartTime | Now required; used with ProjectID to locate the run in SmithDB |
| (not available) | ProjectID | Required—UUID of the project that owns the run |
| (all fields returned by default) | Selects | Field projection; defaults to ["ID"] only; field name constants are uppercase (e.g., RunGetV2ParamsSelectName) |
run_id remains in the URL path. All other parameters are query string values with snake_case names.GET /v2/runs/{run_id} requires a new project_id query param and makes start_time required (previously optional). Both are needed to locate the run in SmithDB.Before (GET /api/v1/runs/{run_id} param) | After (GET /v2/runs/{run_id} param) | Notes |
|---|---|---|
run_id (path) | run_id (path) | Unchanged |
load_child_runs (query) | (removed) | No equivalent |
| (not available) | project_id (query) | Required—UUID of the project that owns the run |
start_time (query) | start_time (query) | Now required; used with project_id to locate the run in SmithDB |
| (all fields returned by default) | selects (query, repeatable) | Field projection; defaults to ["ID"] only; field names are uppercase |
Response fields
- Python
- TypeScript
- Java
- Go
- cURL
Pass SCREAMING_SNAKE_CASE strings to
selects (eg. "ID", "NAME", "STATUS") to control which fields are populated on the returned Run; only selected fields are non-None. Default selects contains only "ID".Before (v1 Run attribute) | After (v2 Run attribute) | Notes |
|---|---|---|
run.id | run.id | Unchanged; returned by default when selects is omitted |
run.name | run.name | Unchanged |
run.run_type | run.run_type | Values are now uppercase Literals: "LLM", "CHAIN", etc. |
run.status | run.status | Values: "SUCCESS", "ERROR", "PENDING" |
run.start_time | run.start_time | Unchanged |
run.end_time | run.end_time | Unchanged |
run.error | run.error | Unchanged |
run.inputs | run.inputs | Unchanged |
run.outputs | run.outputs | Unchanged |
run.tags | run.tags | Unchanged |
run.extra | run.extra | Unchanged |
run.metadata | run.metadata | Unchanged |
run.events | run.events | Unchanged |
run.reference_example_id | run.reference_example_id | Unchanged |
run.trace_id | run.trace_id | Unchanged |
run.dotted_order | run.dotted_order | Unchanged |
run.parent_run_id | (removed) | Use run.parent_run_ids (list of all ancestor UUIDs, root first) |
run.parent_run_ids | run.parent_run_ids | Unchanged |
run.session_id | run.project_id | Renamed; session_id was the project UUID |
run.feedback_stats | run.feedback_stats | Unchanged |
run.app_path | run.app_path | Unchanged |
run.attachments | run.attachments | v2 returns pre-signed download URLs instead of raw bytes |
run.total_tokens | run.total_tokens | Unchanged |
run.prompt_tokens | run.prompt_tokens | Unchanged |
run.completion_tokens | run.completion_tokens | Unchanged |
run.total_cost | run.total_cost | Unchanged |
run.prompt_cost | run.prompt_cost | Unchanged |
run.completion_cost | run.completion_cost | Unchanged |
run.first_token_time | run.first_token_time | Unchanged |
run.latency (property) | run.latency_seconds | Renamed; was a computed timedelta property, now a native float field |
run.in_dataset | run.is_in_dataset | Renamed |
run.child_run_ids | (removed) | No equivalent |
run.child_runs | (removed) | No equivalent |
run.serialized | (removed) | Use run.manifest |
run.manifest_id | (removed) | Use run.manifest |
| (not available) | run.is_root | New |
| (not available) | run.manifest | New: full manifest object (replaces serialized and manifest_id) |
| (not available) | run.error_preview | New: truncated error snippet |
| (not available) | run.inputs_preview | New: truncated inputs preview |
| (not available) | run.outputs_preview | New: truncated outputs preview |
| (not available) | run.thread_id | New: conversation thread UUID |
| (not available) | run.reference_dataset_id | New: dataset UUID for the reference example |
| (not available) | run.share_url | New: public share URL (only set when the run has been shared) |
run.prompt_token_details | run.prompt_token_details.raw | Field now wraps the dict; access .raw to get dict[str, int] (element type unchanged) |
run.completion_token_details | run.completion_token_details.raw | Field now wraps the dict; access .raw to get dict[str, int] (element type unchanged) |
run.prompt_cost_details | run.prompt_cost_details.raw | Field now wraps the dict; access .raw to get dict[str, float] (was dict[str, Decimal]) |
run.completion_cost_details | run.completion_cost_details.raw | Field now wraps the dict; access .raw to get dict[str, float] (was dict[str, Decimal]) |
Pass SCREAMING_SNAKE_CASE strings to
selects (eg. "ID", "NAME", "STATUS") to control which fields are populated on the returned Run. Default selects contains only "ID".Before (v1 Run property) | After (v2 Run property) | Notes |
|---|---|---|
run.id | run.id | Unchanged |
run.name | run.name | Unchanged |
run.runType | run.run_type | Renamed to snake_case; values are now uppercase: "LLM", "CHAIN", etc. |
run.status | run.status | Values: "SUCCESS", "ERROR", "PENDING" |
run.startTime | run.start_time | Renamed to snake_case |
run.endTime | run.end_time | Renamed to snake_case |
run.error | run.error | Unchanged |
run.inputs | run.inputs | Unchanged |
run.outputs | run.outputs | Unchanged |
run.tags | run.tags | Unchanged |
run.extra | run.extra | Unchanged |
| (not available) | run.metadata | New: previously accessed via run.extra.metadata |
run.events | run.events | Unchanged |
run.referenceExampleId | run.reference_example_id | Renamed to snake_case |
run.traceId | run.trace_id | Renamed to snake_case |
run.dottedOrder | run.dotted_order | Renamed to snake_case |
run.parentRunId | (removed) | Use run.parent_run_ids (list of all ancestor UUIDs, root first) |
run.parentRunIds | run.parent_run_ids | Renamed to snake_case |
run.sessionId | run.project_id | Renamed; sessionId was the project UUID |
run.feedbackStats | run.feedback_stats | Renamed to snake_case |
run.appPath | run.app_path | Renamed to snake_case |
run.attachments | run.attachments | v2 returns pre-signed download URLs instead of raw bytes |
run.totalTokens | run.total_tokens | Renamed to snake_case |
run.promptTokens | run.prompt_tokens | Renamed to snake_case |
run.completionTokens | run.completion_tokens | Renamed to snake_case |
run.totalCost | run.total_cost | Renamed to snake_case |
run.promptCost | run.prompt_cost | Renamed to snake_case |
run.completionCost | run.completion_cost | Renamed to snake_case |
run.firstTokenTime | run.first_token_time | Renamed to snake_case |
run.latency | run.latency_seconds | Renamed; was a computed property, now a native number field (seconds) |
run.inDataset | run.is_in_dataset | Renamed |
run.childRunIds | (removed) | No equivalent |
run.childRuns | (removed) | No equivalent |
run.serialized | (removed) | Use run.manifest |
run.manifestId | (removed) | Use run.manifest |
run.shareToken | (removed) | Use run.share_url (full URL, only set when the run has been shared) |
| (not available) | run.is_root | New |
| (not available) | run.manifest | New: full manifest object (replaces serialized and manifestId) |
| (not available) | run.error_preview | New: truncated error snippet |
| (not available) | run.inputs_preview | New: truncated inputs preview |
| (not available) | run.outputs_preview | New: truncated outputs preview |
| (not available) | run.thread_id | New: conversation thread UUID |
| (not available) | run.reference_dataset_id | New: dataset UUID for the reference example |
| (not available) | run.share_url | New: public share URL (only set when the run has been shared) |
| (not available) | run.prompt_token_details | New: per-category prompt token breakdown |
| (not available) | run.completion_token_details | New: per-category completion token breakdown |
| (not available) | run.prompt_cost_details | New: per-category prompt cost breakdown |
| (not available) | run.completion_cost_details | New: per-category completion cost breakdown |
Add
RunRetrieveV2Params.Select values (eg. Select.NAME, Select.STATUS) via .addSelect(...) to control which fields are populated; unselected fields return empty Optional values. selects() defaults to ID only.Before (RunSchema method) | After (Run method) | Notes |
|---|---|---|
run.id() | run.id() | Unchanged |
run.name() | run.name() | Unchanged |
run.runType() | run.runType() | Values are now uppercase: "LLM", "CHAIN", etc. |
run.status() | run.status() | Values: "SUCCESS", "ERROR", "PENDING" |
run.startTime() | run.startTime() | Unchanged |
run.endTime() | run.endTime() | Unchanged |
run.error() | run.error() | Unchanged |
run.inputs() | run.inputs() | Unchanged |
run.outputs() | run.outputs() | Unchanged |
run.tags() | run.tags() | Unchanged |
run.extra() | run.extra() | Unchanged |
run.events() | run.events() | Unchanged |
run.feedbackStats() | run.feedbackStats() | Unchanged |
run.inputsPreview() | run.inputsPreview() | Unchanged |
run.outputsPreview() | run.outputsPreview() | Unchanged |
run.referenceExampleId() | run.referenceExampleId() | Unchanged |
run.traceId() | run.traceId() | Unchanged |
run.dottedOrder() | run.dottedOrder() | Unchanged |
run.parentRunId() | (removed) | Use run.parentRunIds() (list of all ancestor UUIDs, root first) |
run.parentRunIds() | run.parentRunIds() | Unchanged |
run.sessionId() | run.projectId() | Renamed; sessionId() returned the project UUID |
run.appPath() | run.appPath() | Unchanged |
run.firstTokenTime() | run.firstTokenTime() | Unchanged |
run.totalTokens() | run.totalTokens() | Unchanged |
run.promptTokens() | run.promptTokens() | Unchanged |
run.completionTokens() | run.completionTokens() | Unchanged |
run.totalCost() | run.totalCost() | Return type changed from Optional<String> to Optional<Double> |
run.promptCost() | run.promptCost() | Return type changed from Optional<String> to Optional<Double> |
run.completionCost() | run.completionCost() | Return type changed from Optional<String> to Optional<Double> |
run.promptTokenDetails() | run.promptTokenDetails() | Unchanged |
run.completionTokenDetails() | run.completionTokenDetails() | Unchanged |
run.promptCostDetails() | run.promptCostDetails() | Unchanged |
run.completionCostDetails() | run.completionCostDetails() | Unchanged |
run.priceModelId() | run.priceModelId() | Unchanged |
run.inDataset() | run.isInDataset() | Renamed |
run.referenceDatasetId() | run.referenceDatasetId() | Unchanged |
run.threadId() | run.threadId() | Unchanged |
run.shareToken() | (removed) | Use run.shareUrl() (full URL, only set when the run has been shared) |
run.childRunIds() | (removed) | No equivalent |
run.directChildRunIds() | (removed) | No equivalent |
run.serialized() | (removed) | Use run.manifest() |
run.manifestId() | (removed) | Use run.manifest() |
run.messages() | (removed) | No equivalent |
run.executionOrder() | (removed) | No equivalent |
run.lastQueuedAt() | (removed) | No equivalent |
run.traceFirstReceivedAt() | (removed) | No equivalent |
run.traceMaxStartTime() | (removed) | No equivalent |
run.traceMinStartTime() | (removed) | No equivalent |
run.traceTier() | (removed) | No equivalent |
run.traceUpgrade() | (removed) | No equivalent |
run.ttlSeconds() | (removed) | No equivalent |
| (not available) | run.attachments() | New: pre-signed download URLs for attachments (replaces S3 URL fields) |
| (not available) | run.latencySeconds() | New: wall-clock duration in seconds |
| (not available) | run.isRoot() | New |
| (not available) | run.errorPreview() | New: truncated error snippet |
| (not available) | run.manifest() | New: full manifest, typed as Optional<Manifest> (replaces serialized() and manifestId()) |
| (not available) | run.metadata() | New: metadata, typed as Optional<Metadata> (was derived from extra.metadata) |
| (not available) | run.shareUrl() | New: public share URL (only set when the run has been shared) |
| (not available) | run.threadEvaluationTime() | New |
Pass
RunGetV2ParamsSelect constants (eg. RunGetV2ParamsSelectName, RunGetV2ParamsSelectStatus) to Selects to control which fields are populated; unselected fields are zero-valued on the returned struct. Selects defaults to ID only.Before (RunSchema field) | After (Run field) | Notes |
|---|---|---|
run.ID | run.ID | Unchanged |
run.Name | run.Name | Unchanged |
run.RunType | run.RunType | Values changed to uppercase: "LLM", "CHAIN", etc. |
run.Status | run.Status | Values: "SUCCESS", "ERROR", "PENDING" |
run.TraceID | run.TraceID | Unchanged |
run.DottedOrder | run.DottedOrder | Unchanged |
run.AppPath | run.AppPath | Unchanged |
run.StartTime | run.StartTime | Unchanged |
run.EndTime | run.EndTime | Unchanged |
run.Error | run.Error | Unchanged |
run.Events | run.Events | Unchanged; element type is now RunEvent (was map[string]interface{}) |
run.Extra | run.Extra | Unchanged; type is now interface{} (was map[string]interface{}) |
run.FeedbackStats | run.FeedbackStats | Unchanged; element type is now RunFeedbackStat |
run.FirstTokenTime | run.FirstTokenTime | Unchanged |
run.Inputs | run.Inputs | Unchanged; type is now interface{} (was map[string]interface{}) |
run.InputsPreview | run.InputsPreview | Unchanged |
run.Outputs | run.Outputs | Unchanged; type is now interface{} (was map[string]interface{}) |
run.OutputsPreview | run.OutputsPreview | Unchanged |
run.ParentRunIDs | run.ParentRunIDs | Unchanged |
run.PriceModelID | run.PriceModelID | Unchanged |
run.PromptCost | run.PromptCost | Unchanged |
run.PromptCostDetails | run.PromptCostDetails.Raw | Field now wraps the map; access .Raw to get map[string]float64 (was map[string]string) |
run.PromptTokenDetails | run.PromptTokenDetails.Raw | Field now wraps the map; access .Raw to get map[string]int64 (element type unchanged) |
run.PromptTokens | run.PromptTokens | Unchanged |
run.CompletionCost | run.CompletionCost | Unchanged |
run.CompletionCostDetails | run.CompletionCostDetails.Raw | Field now wraps the map; access .Raw to get map[string]float64 (was map[string]string) |
run.CompletionTokenDetails | run.CompletionTokenDetails.Raw | Field now wraps the map; access .Raw to get map[string]int64 (element type unchanged) |
run.CompletionTokens | run.CompletionTokens | Unchanged |
run.TotalCost | run.TotalCost | Unchanged |
run.TotalTokens | run.TotalTokens | Unchanged |
run.ReferenceDatasetID | run.ReferenceDatasetID | Unchanged |
run.ReferenceExampleID | run.ReferenceExampleID | Unchanged |
run.Tags | run.Tags | Unchanged |
run.ThreadID | run.ThreadID | Unchanged |
run.SessionID | run.ProjectID | Renamed |
run.InDataset | run.IsInDataset | Renamed |
run.ChildRunIDs | (removed) | No equivalent |
run.DirectChildRunIDs | (removed) | No equivalent |
run.ExecutionOrder | (removed) | No equivalent |
run.InputsS3URLs | (removed) | Internal storage URL; not exposed in v2 |
run.LastQueuedAt | (removed) | No equivalent |
run.ManifestID | (removed) | Use run.Manifest |
run.ManifestS3ID | (removed) | Internal storage URL; not exposed in v2 |
run.Messages | (removed) | No equivalent |
run.OutputsS3URLs | (removed) | Internal storage URL; not exposed in v2 |
run.ParentRunID | (removed) | Use run.ParentRunIDs |
run.S3URLs | (removed) | Internal storage URL; not exposed in v2 |
run.Serialized | (removed) | Use run.Manifest |
run.ShareToken | (removed) | Use run.ShareURL |
run.TraceFirstReceivedAt | (removed) | No equivalent |
run.TraceMaxStartTime | (removed) | No equivalent |
run.TraceMinStartTime | (removed) | No equivalent |
run.TraceTier | (removed) | No equivalent |
run.TraceUpgrade | (removed) | No equivalent |
run.TtlSeconds | (removed) | No equivalent |
| (not available) | run.Attachments | New: maps attachment filename to pre-signed download URL |
| (not available) | run.ErrorPreview | New: truncated error snippet |
| (not available) | run.IsRoot | New |
| (not available) | run.LatencySeconds | New: wall-clock duration in seconds |
| (not available) | run.Manifest | New: full manifest object (replaces Serialized and ManifestID) |
| (not available) | run.Metadata | New: arbitrary user-defined JSON metadata |
| (not available) | run.ShareURL | New: public share URL (only set when the run has been shared) |
| (not available) | run.ThreadEvaluationTime | New |
Pass SCREAMING_SNAKE_CASE strings as repeated
selects query parameters (eg. selects=NAME&selects=STATUS) to control which fields are populated. Default selects contains only "ID".| Before (v1 response field) | After (v2 response field) | Notes |
|---|---|---|
id | id | Unchanged |
name | name | Unchanged |
run_type | run_type | Values changed to uppercase: "LLM", "CHAIN", etc. |
status | status | Values: "SUCCESS", "ERROR", "PENDING" |
trace_id | trace_id | Unchanged |
dotted_order | dotted_order | Unchanged |
app_path | app_path | Unchanged |
start_time | start_time | Unchanged |
end_time | end_time | Unchanged |
error | error | Unchanged |
events | events | Unchanged |
extra | extra | Unchanged |
feedback_stats | feedback_stats | Unchanged |
first_token_time | first_token_time | Unchanged |
inputs | inputs | Unchanged |
inputs_preview | inputs_preview | Unchanged |
outputs | outputs | Unchanged |
outputs_preview | outputs_preview | Unchanged |
parent_run_ids | parent_run_ids | Unchanged |
price_model_id | price_model_id | Unchanged |
prompt_cost | prompt_cost | Unchanged |
prompt_cost_details | prompt_cost_details.raw | Field now wraps the object; read .raw for the same {category: cost} mapping, now with numeric values (was strings) |
prompt_token_details | prompt_token_details.raw | Field now wraps the object; read .raw for the same {category: count} mapping (values unchanged) |
prompt_tokens | prompt_tokens | Unchanged |
completion_cost | completion_cost | Unchanged |
completion_cost_details | completion_cost_details.raw | Field now wraps the object; read .raw for the same {category: cost} mapping, now with numeric values (was strings) |
completion_token_details | completion_token_details.raw | Field now wraps the object; read .raw for the same {category: count} mapping (values unchanged) |
completion_tokens | completion_tokens | Unchanged |
total_cost | total_cost | Unchanged |
total_tokens | total_tokens | Unchanged |
reference_dataset_id | reference_dataset_id | Unchanged |
reference_example_id | reference_example_id | Unchanged |
tags | tags | Unchanged |
thread_id | thread_id | Unchanged |
session_id | project_id | Renamed |
in_dataset | is_in_dataset | Renamed |
child_run_ids | (removed) | No equivalent |
direct_child_run_ids | (removed) | No equivalent |
execution_order | (removed) | No equivalent |
inputs_s3_urls | (removed) | Internal storage URL; not exposed in v2 |
last_queued_at | (removed) | No equivalent |
manifest_id | (removed) | Use manifest |
manifest_s3_id | (removed) | Internal storage URL; not exposed in v2 |
messages | (removed) | No equivalent |
outputs_s3_urls | (removed) | Internal storage URL; not exposed in v2 |
parent_run_id | (removed) | Use parent_run_ids |
s3_urls | (removed) | Internal storage URL; not exposed in v2 |
serialized | (removed) | Use manifest |
share_token | (removed) | Use share_url |
trace_first_received_at | (removed) | No equivalent |
trace_max_start_time | (removed) | No equivalent |
trace_min_start_time | (removed) | No equivalent |
trace_tier | (removed) | No equivalent |
trace_upgrade | (removed) | No equivalent |
ttl_seconds | (removed) | No equivalent |
| (not available) | attachments | New: maps attachment filename to pre-signed download URL |
| (not available) | error_preview | New: truncated error snippet |
| (not available) | is_root | New |
| (not available) | latency_seconds | New: wall-clock duration in seconds |
| (not available) | manifest | New: full manifest object (replaces serialized and manifest_id) |
| (not available) | metadata | New: previously nested under extra.metadata |
| (not available) | share_url | New: public share URL (only set when the run has been shared) |
| (not available) | thread_evaluation_time | New |
Examples
Fetch a single run by ID
- Python
- TypeScript
- Java
- Go
- cURL
runs.retrieve requires two additional parameters that read_run did not need: project_id (UUID) and start_time. Both are required by the SmithDB storage model to locate a run efficiently. Resolve the project UUID via client.aread_project() first.- Before
- After
Before
from langsmith import Client
client = Client()
run_id = "<run-id>"
run = client.read_run(run_id)
After
import asyncio
from langsmith import Client
async def main():
client = Client()
project = await client.aread_project(project_name="default")
run_id = "<run-id>"
start_time = "2026-06-01T12:00:00Z"
run = await client.runs.retrieve(
run_id=run_id,
project_id=str(project.id),
start_time=start_time,
)
asyncio.run(main())
client.runs.retrieve requires two additional parameters that readRun did not need: project_id (UUID) and start_time. Both are required by the SmithDB storage model to locate a run efficiently. Resolve the project UUID via client.readProject() first.- Before
- After
Before
import { Client } from "langsmith";
const client = new Client();
let runId = "<run-id>";
await client.readRun(runId);
After
import { Client } from "langsmith";
const client = new Client();
const project = await client.readProject({ projectName: "default" });
let runId = "<run-id>";
let startTime = "2026-06-01T12:00:00Z";
await client.runs.retrieve(runId, {
project_id: project.id,
start_time: startTime,
});
retrieveV2() requires two additional parameters that client.runs().retrieve() did not need: projectId() (UUID) and startTime(). Both are required by the SmithDB storage model to locate a run efficiently. Resolve the project UUID via client.sessions().list() first.- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
var runId = "<run-id>"
client.runs().retrieve(runId)
After
import java.time.OffsetDateTime
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunRetrieveV2Params
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
var runId = "<run-id>"
var startTime = "<run-start-time-rfc3339>"
client.runs().retrieveV2(
runId,
RunRetrieveV2Params.builder()
.projectId(project.id())
.startTime(OffsetDateTime.parse(startTime))
.build()
)
GetV2() requires two additional parameters that client.Runs.Get() did not need: ProjectID (UUID) and StartTime. Both are required by the SmithDB storage model to locate a run efficiently. Resolve the project UUID via client.Sessions.List() first.- Before
- After
Before
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
runID := "<run-id>"
run, err := client.Runs.Get(ctx, runID, langsmith.RunGetParams{})
After
package main
import (
"context"
"time"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
runID := "<run-id>"
startTime := time.Date(2026, 6, 1, 12, 0, 0, 0, time.UTC)
projectID := "<project-id>"
run, err := client.Runs.GetV2(ctx, runID, langsmith.RunGetV2Params{
ProjectID: langsmith.F(projectID),
StartTime: langsmith.F(startTime),
})
GET /v2/runs/{run_id} requires two additional query parameters that GET /api/v1/runs/{run_id} did not need: project_id (UUID) and start_time. Both are required by the SmithDB storage model to locate a run efficiently. Resolve the project UUID via a GET /api/v1/sessions request first.- Before
- After
RUN_ID="<run-id>"
curl "https://api.smith.langchain.com/api/v1/runs/$RUN_ID" \
-H "x-api-key: $LANGSMITH_API_KEY"
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
RUN_ID="<run-id>"
START_TIME="2025-01-01T12:00:00Z"
curl "https://api.smith.langchain.com/v2/runs/$RUN_ID?project_id=$PROJECT_ID&start_time=$START_TIME" \
-H "x-api-key: $LANGSMITH_API_KEY"
Selecting fields
- Python
- TypeScript
- Java
- Go
- cURL
read_run returns a full run object with no selection needed. runs.retrieve returns only id by default—pass selects=[...] to request more.- Before
- After
Before
from langsmith import Client
client = Client()
run_id = "<run-id>"
run = client.read_run(run_id=run_id)
print(run.name, run.status, run.total_tokens)
After
import asyncio
from langsmith import Client
async def main():
client = Client()
project = await client.aread_project(project_name="default")
run_id = "<run-id>"
start_time = "2026-06-01T12:00:00Z"
run = await client.runs.retrieve(
run_id=run_id,
project_id=str(project.id),
start_time=start_time,
selects=["NAME", "STATUS", "TOTAL_TOKENS"],
)
print(run.name, run.status, run.total_tokens)
asyncio.run(main())
readRun returns a full run object with no selection needed. client.runs.retrieve returns only id by default—pass selects: [...] to request more.- Before
- After
Before
import { Client } from "langsmith";
const client = new Client();
let runId = "<run-id>";
const retrievedRun = await client.readRun(runId);
console.log(retrievedRun.name, retrievedRun.status, retrievedRun.total_tokens);
After
import { Client } from "langsmith";
const client = new Client();
let runId = "<run-id>";
let startTime = "2026-06-01T12:00:00Z";
let projectId = "<project-id>";
const retrievedRun = await client.runs.retrieve(runId, {
project_id: projectId,
start_time: startTime,
selects: ["NAME", "STATUS", "TOTAL_TOKENS"],
});
console.log(retrievedRun.name, retrievedRun.status, retrievedRun.total_tokens);
.retrieve() returns a full run object with no selection needed. .retrieveV2() returns only id by default—call .addSelect(...) for each field you need.- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
var runId = "<run-id>"
val run = client.runs().retrieve(runId)
println("${run.name()} ${run.status()} ${run.totalTokens()}")
After
import java.time.OffsetDateTime
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.runs.RunRetrieveV2Params
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
var runId = "<run-id>"
var startTime = "<run-start-time-rfc3339>"
val run = client.runs().retrieveV2(
runId,
RunRetrieveV2Params.builder()
.projectId(project.id())
.startTime(OffsetDateTime.parse(startTime))
.addSelect(RunRetrieveV2Params.Select.NAME)
.addSelect(RunRetrieveV2Params.Select.STATUS)
.addSelect(RunRetrieveV2Params.Select.TOTAL_TOKENS)
.build()
)
println("${run.name()} ${run.status()} ${run.totalTokens()}")
Get returns a full run struct with no selection needed. GetV2 returns only ID by default—pass Selects with the fields you need.- Before
- After
Before
package main
import (
"context"
"fmt"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
runID := "<run-id>"
run, err := client.Runs.Get(ctx, runID, langsmith.RunGetParams{})
fmt.Println(run.Name, run.Status, run.TotalTokens)
After
package main
import (
"context"
"fmt"
"time"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
runID := "<run-id>"
startTime := time.Date(2026, 6, 1, 12, 0, 0, 0, time.UTC)
projectID := "<project-id>"
run, err := client.Runs.GetV2(ctx, runID, langsmith.RunGetV2Params{
ProjectID: langsmith.F(projectID),
StartTime: langsmith.F(startTime),
Selects: langsmith.F([]langsmith.RunGetV2ParamsSelect{
langsmith.RunGetV2ParamsSelectName,
langsmith.RunGetV2ParamsSelectStatus,
langsmith.RunGetV2ParamsSelectTotalTokens,
}),
})
fmt.Println(run.Name, run.Status, run.TotalTokens)
GET /api/v1/runs/{run_id} returns a full run object with no selection needed. GET /v2/runs/{run_id} returns only id by default—pass selects query parameters for the fields you need.- Before
- After
RUN_ID="<run-id>"
curl "https://api.smith.langchain.com/api/v1/runs/$RUN_ID" \
-H "x-api-key: $LANGSMITH_API_KEY"
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
RUN_ID="<run-id>"
START_TIME="2026-06-01T12:00:00Z"
curl "https://api.smith.langchain.com/v2/runs/$RUN_ID?project_id=$PROJECT_ID&start_time=$START_TIME&selects=NAME&selects=STATUS&selects=TOTAL_TOKENS" \
-H "x-api-key: $LANGSMITH_API_KEY"
Handle a not-found run
- Python
- TypeScript
- Java
- Go
- cURL
read_run raised LangSmithNotFoundError from langsmith.utils for a missing run. runs.retrieve raises NotFoundError from langsmith instead.- Before
- After
Before
from langsmith import Client
from langsmith.utils import LangSmithNotFoundError
client = Client()
run_id = "<run-id>"
try:
run = client.read_run(run_id)
except LangSmithNotFoundError:
print(f"Run {run_id} not found")
After
import asyncio
from langsmith import Client
from langsmith import NotFoundError
async def main():
client = Client()
project = await client.aread_project(project_name="default")
run_id = "<run-id>"
start_time = "2026-06-01T12:00:00Z"
try:
run = await client.runs.retrieve(
run_id=run_id,
project_id=str(project.id),
start_time=start_time,
)
except NotFoundError:
print(f"Run {run_id} not found")
asyncio.run(main())
client.runs.retrieve raises NotFoundError for a missing run.- Before
- After
Before
import { Client } from "langsmith";
const client = new Client();
let runId = "<run-id>";
try {
await client.readRun(runId);
} catch (e: any) {
if (e?.status === 404) {
console.log(`Run ${runId} not found`);
}
}
After
import { Client, NotFoundError } from "langsmith";
const client = new Client();
const project = await client.readProject({ projectName: "default" });
let runId = "<run-id>";
const startTime = "2026-06-01T12:00:00Z";
try {
await client.runs.retrieve(runId, {
project_id: project.id,
start_time: startTime,
});
} catch (e) {
if (e instanceof NotFoundError) {
console.log(`Run ${runId} not found`);
}
}
.retrieve() and .retrieveV2() both raise com.langchain.smith.errors.NotFoundException—unchanged, since the Java SDK was already Stainless-generated before SmithDB.- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.errors.NotFoundException
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
var runId = "<run-id>"
try {
client.runs().retrieve(runId)
} catch (e: NotFoundException) {
println("Run $runId not found")
}
After
import java.time.OffsetDateTime
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.errors.NotFoundException
import com.langchain.smith.models.runs.RunRetrieveV2Params
import com.langchain.smith.models.sessions.SessionListParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val project = client.sessions().list(
SessionListParams.builder().name("default").limit(1L).build()
).items().first()
var runId = "<run-id>"
var startTime = "<run-start-time-rfc3339>"
try {
client.runs().retrieveV2(
runId,
RunRetrieveV2Params.builder()
.projectId(project.id())
.startTime(OffsetDateTime.parse(startTime))
.build()
)
} catch (e: NotFoundException) {
println("Run $runId not found")
}
Get and GetV2 both return a *langsmith.Error you can inspect with errors.As—unchanged, since the Go SDK was already Stainless-generated before SmithDB. Check StatusCode for 404.- Before
- After
Before
package main
import (
"context"
"errors"
"fmt"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
runID := "<run-id>"
_, err := client.Runs.Get(ctx, runID, langsmith.RunGetParams{})
if err != nil {
var apiErr *langsmith.Error
if errors.As(err, &apiErr) && apiErr.StatusCode == 404 {
fmt.Printf("Run %s not found\n", runID)
} else {
panic(err)
}
}
After
package main
import (
"context"
"errors"
"fmt"
"time"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
runID := "<run-id>"
startTime := time.Date(2026, 6, 1, 12, 0, 0, 0, time.UTC)
projectID := "<project-id>"
_, err := client.Runs.GetV2(ctx, runID, langsmith.RunGetV2Params{
ProjectID: langsmith.F(projectID),
StartTime: langsmith.F(startTime),
})
if err != nil {
var apiErr *langsmith.Error
if errors.As(err, &apiErr) && apiErr.StatusCode == 404 {
fmt.Printf("Run %s not found\n", runID)
} else {
panic(err)
}
}
Both
GET /api/v1/runs/{run_id} and GET /v2/runs/{run_id} return HTTP 404 for a missing run. Check the response status code.- Before
- After
RUN_ID="<run-id>"
HTTP_STATUS=$(curl -s -o /dev/null -w "%{http_code}" \
"https://api.smith.langchain.com/api/v1/runs/$RUN_ID" \
-H "x-api-key: $LANGSMITH_API_KEY")
if [ "$HTTP_STATUS" = "404" ]; then
echo "Run $RUN_ID not found"
fi
PROJECT_ID=$(curl -s "https://api.smith.langchain.com/api/v1/sessions?name=default&limit=1" \
-H "x-api-key: $LANGSMITH_API_KEY" | jq -r '.[0].id')
RUN_ID="<run-id>"
START_TIME="2025-01-01T12:00:00Z"
HTTP_STATUS=$(curl -s -o /dev/null -w "%{http_code}" \
"https://api.smith.langchain.com/v2/runs/$RUN_ID?project_id=$PROJECT_ID&start_time=$START_TIME" \
-H "x-api-key: $LANGSMITH_API_KEY")
if [ "$HTTP_STATUS" = "404" ]; then
echo "Run $RUN_ID not found"
fi
Dataset experiment runs: query
Query dataset examples together with the experiment runs recorded against each example. Accepts one or moreexperiment_ids so you can view runs from multiple experiments side by side; results are returned as a cursor-paginated page.
Main changes
Method name
- Python
- TypeScript
- Java
- Go
- cURL
| Before | After |
|---|---|
client.get_experiment_results() | client.datasets.experiment_runs.query() |
| Before | After |
|---|---|
(no legacy public Client method) | client.datasets.experimentRuns.query() |
| Before | After |
|---|---|
client.datasets().runs().query() | client.datasets().experimentRuns().query() |
| Before | After |
|---|---|
client.Datasets.Runs.Query() | client.Datasets.ExperimentRuns.Query() |
| Before | After |
|---|---|
POST /api/v1/datasets/{dataset_id}/runs | POST /v2/datasets/{dataset_id}/experiment-runs |
Query parameters
- Python
- TypeScript
- Java
- Go
- cURL
experiment_ids is required and replaces session_ids. Values are still experiment tracing-project UUIDs—if you only know the experiment’s name, resolve it first: client.read_project(project_name="my-experiment").id, or await client.aread_project(project_name="my-experiment") in async code.Before (get_experiment_results) | After (datasets.experiment_runs.query) | Notes |
|---|---|---|
project_id | experiment_ids | get_experiment_results accepted one project/experiment; the new method accepts a required non-empty list |
limit | (removed) | Use page_size for per-request batch size |
| (not available) | page_size | Per-request result count (default 20, max 100) |
| (handled internally) | cursor | Pass the previous page’s next_cursor to fetch the next page |
preview | selects | Omitted selects returns only run IDs; use INPUTS_PREVIEW and OUTPUTS_PREVIEW for previews, or INPUTS and OUTPUTS for full payloads |
| (not exposed) | sort | Use {by, order} for feedback-score sorting |
filters | filters | Unchanged; maps experiment UUID strings to filter expressions |
comparative_experiment_id | comparative_experiment_id | Unchanged |
| (not exposed) | example_ids | Optional example UUID filter, max 1000 |
experiment_ids is required and replaces session_ids. Values are still experiment tracing-project UUIDs—if you only know the experiment’s name, resolve it first: (await client.readProject({ projectName: "my-experiment" })).id.| Before | After (datasets.experimentRuns.query) | Notes |
|---|---|---|
(no legacy public Client method) | experiment_ids | Required and non-empty |
(no legacy public Client method) | page_size | Defaults to 20, max 100 |
(no legacy public Client method) | cursor | Pass the previous page’s next_cursor instead of a numeric offset |
(no legacy public Client method) | selects | Omitted selects returns only run IDs; use INPUTS_PREVIEW and OUTPUTS_PREVIEW for previews, or INPUTS and OUTPUTS for full payloads |
(no legacy public Client method) | sort | Use { by, order } for feedback-score sorting |
(no legacy public Client method) | filters | Maps experiment UUID strings to filter expressions |
(no legacy public Client method) | comparative_experiment_id | Scopes pairwise-annotation feedback |
(no legacy public Client method) | example_ids | Optional example UUID filter, max 1000 |
experimentIds() is required and replaces sessionIds(). Values are still experiment tracing-project UUIDs—if you only know the experiment’s name, resolve it first: client.sessions().list(SessionListParams.builder().name("my-experiment").build()).items().first().id().Before (RunQueryParams) | After (ExperimentRunQueryParams) | Notes |
|---|---|---|
sessionIds() | experimentIds() | Renamed; required and non-empty |
limit() | (removed) | Use pageSize() for per-request batch size |
| (not available) | pageSize() | Per-request result count (default 20, max 100) |
offset() | cursor() | Pass the previous page’s nextCursor() instead of a numeric offset |
preview() | selects() | Omitted selects return only run IDs; add Select.INPUTS_PREVIEW and Select.OUTPUTS_PREVIEW for previews |
sortParams() | sort() | Shape changed from sortBy() / sortOrder() to by() / order() |
filters() | filters() | Unchanged |
comparativeExperimentId() | comparativeExperimentId() | Unchanged |
exampleIds() | exampleIds() | Unchanged, max 1000 |
format() | (removed) | The new endpoint returns JSON only |
includeAnnotatorDetail() | (removed) | No new JSON equivalent |
ExperimentIDs is required and replaces SessionIDs. Values are still experiment tracing-project UUIDs—if you only know the experiment’s name, resolve it first: list sessions filtered by Name and take the first result’s ID.Before (DatasetRunQueryParams) | After (DatasetExperimentRunQueryParams) | Notes |
|---|---|---|
SessionIDs | ExperimentIDs | Renamed; required and non-empty |
Limit | (removed) | Use PageSize for per-request batch size |
| (not available) | PageSize | Per-request result count (default 20, max 100) |
Offset | Cursor | Pass the previous page’s NextCursor instead of a numeric offset |
Preview | Selects | Omitted selects return only run IDs; use InputsPreview and OutputsPreview select constants for previews |
SortParams | Sort | Shape changed from SortBy / SortOrder to By / Order |
Filters | Filters | Unchanged |
ComparativeExperimentID | ComparativeExperimentID | Unchanged |
ExampleIDs | ExampleIDs | Unchanged, max 1000 |
Format | (removed) | The new endpoint returns JSON only |
IncludeAnnotatorDetail | (removed) | No new JSON equivalent |
experiment_ids is required and replaces session_ids. Values are still experiment tracing-project UUIDs—if you only know the experiment’s name, resolve it first: GET /api/v1/sessions?name=my-experiment and take .[0].id.Before (POST /api/v1/datasets/{dataset_id}/runs body) | After (POST /v2/datasets/{dataset_id}/experiment-runs body) | Notes |
|---|---|---|
session_ids | experiment_ids | Renamed; required and non-empty |
limit | (removed) | Use page_size for per-request batch size |
| (not available) | page_size | Per-request result count (default 20, max 100) |
offset | cursor | Pass the previous page’s next_cursor instead of a numeric offset |
preview | selects | Omitted selects returns only run IDs; use INPUTS_PREVIEW and OUTPUTS_PREVIEW for previews, or INPUTS and OUTPUTS for full payloads |
sort_params | sort | Shape changed from {sort_by, sort_order} to {by, order} |
filters | filters | Unchanged; maps experiment UUID strings to filter expressions |
comparative_experiment_id | comparative_experiment_id | Unchanged |
example_ids | example_ids | Unchanged, max 1000 |
format=csv | (removed) | The new endpoint returns JSON only |
include_annotator_detail | (removed) | No new JSON equivalent |
Response fields
Each page item is a dataset example paired with the runs produced for it—not a bareRun. Its runs field holds the same Run objects returned by Querying runs; see that section for the per-run fields. The tables below describe the rest of the item: the example fields alongside runs.
- Python
- TypeScript
- Java
- Go
- cURL
get_experiment_results returned experiment results with an examples_with_runs iterator. datasets.experiment_runs.query returns a paginated page object (page.items, page.next_cursor); each item has:| Field | Notes |
|---|---|
id | Dataset example UUID |
dataset_id | Parent dataset UUID |
name | Example name, if set |
created_at / modified_at | Example timestamps |
inputs / outputs | Example input and reference-output payloads |
metadata | Example metadata |
source_run_id | Run UUID the example was created from, if any |
attachment_urls | Pre-signed download URL per attachment name |
runs | This example’s runs—see Querying runs |
The legacy dataset runs endpoint was not exposed on the public TypeScript
Client. datasets.experimentRuns.query returns a paginated page (page.getPaginatedItems(), page.next_cursor); each item has:| Field | Notes |
|---|---|
id | Dataset example UUID |
dataset_id | Parent dataset UUID |
name | Example name, if set |
created_at / modified_at | Example timestamps |
inputs / outputs | Example input and reference-output payloads |
metadata | Example metadata |
source_run_id | Run UUID the example was created from, if any |
attachment_urls | Pre-signed download URL per attachment name |
runs | This example’s runs—see Querying runs |
runs().query returned an optional list. experimentRuns().query returns a page object (items(), nextCursor()); each item has:| Field | Notes |
|---|---|
id() | Dataset example UUID |
datasetId() | Parent dataset UUID |
name() | Example name, if set |
createdAt() / modifiedAt() | Example timestamps |
inputs() / outputs() | Example input and reference-output payloads |
metadata() | Example metadata |
sourceRunId() | Run UUID the example was created from, if any |
attachmentUrls() | Pre-signed download URL per attachment name |
runs() | This example’s runs—see Querying runs |
Datasets.Runs.Query returned a slice pointer. Datasets.ExperimentRuns.Query returns an ItemsCursorPostPagination (Items, NextCursor); each item has:| Field | Notes |
|---|---|
ID | Dataset example UUID |
DatasetID | Parent dataset UUID |
Name | Example name, if set |
CreatedAt / ModifiedAt | Example timestamps |
Inputs / Outputs | Example input and reference-output payloads |
Metadata | Example metadata |
SourceRunID | Run UUID the example was created from, if any |
AttachmentURLs | Pre-signed download URL per attachment name |
Runs | This example’s runs—see Querying runs |
POST /api/v1/datasets/{dataset_id}/runs returned a JSON array. POST /v2/datasets/{dataset_id}/experiment-runs returns { "items": [...], "next_cursor": "..." }; each item has:| Field | Notes |
|---|---|
id | Dataset example UUID |
dataset_id | Parent dataset UUID |
name | Example name, if set |
created_at / modified_at | Example timestamps |
inputs / outputs | Example input and reference-output payloads |
metadata | Example metadata |
source_run_id | Run UUID the example was created from, if any |
attachment_urls | Pre-signed download URL per attachment name |
runs | This example’s runs—see Querying runs |
Examples
Query experiment runs and request preview fields
- Python
- TypeScript
- Java
- Go
- cURL
preview=True returned truncated inputs/outputs automatically. In the new API, request that explicitly: pass INPUTS_PREVIEW and OUTPUTS_PREVIEW in selects for the same truncated shape, or INPUTS/OUTPUTS for the untruncated values. Omitting selects returns only id.- Before
- After
Before
from langsmith import Client
client = Client()
experiment_id = client.read_project(project_name=experiment_name).id
results = client.get_experiment_results(
project_id=experiment_id,
limit=20,
preview=True,
)
examples_with_runs = list(results["examples_with_runs"])
After
from langsmith import Client
import asyncio
async def main():
client = Client()
experiment_id = client.read_project(project_name=experiment_name).id
page = await client.datasets.experiment_runs.query(
str(dataset_id),
experiment_ids=[str(experiment_id)],
page_size=20,
selects=["ID", "NAME", "STATUS", "INPUTS_PREVIEW", "OUTPUTS_PREVIEW"],
)
return page.items
examples_with_runs = asyncio.run(main())
The new TypeScript SDK method exposes the experiment-runs query endpoint. The legacy direct endpoint request shape is shown in the cURL tab. Pass
INPUTS_PREVIEW and OUTPUTS_PREVIEW in selects for truncated inputs/outputs, or INPUTS/OUTPUTS for the untruncated values. Omitting selects returns only id.- Before
- After
Before
// The legacy dataset runs endpoint was not exposed on the public TypeScript Client.
// Use the cURL example for the old request body shape.
After
import { Client } from "langsmith";
const client = new Client();
const experimentId = (await client.readProject({ projectName: experimentName })).id;
const page = await client.datasets.experimentRuns.query(datasetId, {
experiment_ids: [experimentId],
page_size: 20,
selects: ["ID", "NAME", "STATUS", "INPUTS_PREVIEW", "OUTPUTS_PREVIEW"],
});
const examplesWithRuns = page.getPaginatedItems();
preview(true) returned truncated inputs/outputs automatically. In the new API, request that explicitly: add Select.INPUTS_PREVIEW and Select.OUTPUTS_PREVIEW for the same truncated shape, or Select.INPUTS/Select.OUTPUTS for the untruncated values. Omitting selects returns only id.- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.datasets.runs.RunQueryParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val examplesWithRuns = client.datasets().runs().query(
datasetId,
RunQueryParams.builder()
.addSessionId(experimentId)
.limit(20L)
.preview(true)
.build()
)
After
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.datasets.experimentruns.ExperimentRunQueryParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val page = client.datasets().experimentRuns().query(
datasetId,
ExperimentRunQueryParams.builder()
.addExperimentId(experimentId)
.pageSize(20L)
.addSelect(ExperimentRunQueryParams.Select.ID)
.addSelect(ExperimentRunQueryParams.Select.NAME)
.addSelect(ExperimentRunQueryParams.Select.STATUS)
.addSelect(ExperimentRunQueryParams.Select.INPUTS_PREVIEW)
.addSelect(ExperimentRunQueryParams.Select.OUTPUTS_PREVIEW)
.build()
)
val examplesWithRuns = page.items()
Preview: true returned truncated inputs/outputs automatically. In the new API, request that explicitly: add the InputsPreview and OutputsPreview select constants for the same truncated shape, or Inputs/Outputs for the untruncated values. Omitting selects returns only ID.- Before
- After
Before
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
examplesWithRuns, err := client.Datasets.Runs.Query(ctx, datasetID, langsmith.DatasetRunQueryParams{
SessionIDs: langsmith.F([]string{experimentID}),
Limit: langsmith.F(int64(20)),
Preview: langsmith.F(true),
})
After
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
page, err := client.Datasets.ExperimentRuns.Query(ctx, datasetID, langsmith.DatasetExperimentRunQueryParams{
ExperimentIDs: langsmith.F([]string{experimentID}),
PageSize: langsmith.F(int64(20)),
Selects: langsmith.F([]langsmith.DatasetExperimentRunQueryParamsSelect{
langsmith.DatasetExperimentRunQueryParamsSelectID,
langsmith.DatasetExperimentRunQueryParamsSelectName,
langsmith.DatasetExperimentRunQueryParamsSelectStatus,
langsmith.DatasetExperimentRunQueryParamsSelectInputsPreview,
langsmith.DatasetExperimentRunQueryParamsSelectOutputsPreview,
}),
})
preview: true returned truncated inputs/outputs automatically. In the new API, request that explicitly: pass INPUTS_PREVIEW and OUTPUTS_PREVIEW in selects for the same truncated shape, or INPUTS/OUTPUTS for the untruncated values. Omitting selects returns only id.- Before
- After
Before
curl -X POST "https://api.smith.langchain.com/api/v1/datasets/$DATASET_ID/runs" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg eid "$EXPERIMENT_ID" '{
"session_ids": [$eid],
"limit": 20,
"preview": true
}')"
After
curl -X POST "https://api.smith.langchain.com/v2/datasets/$DATASET_ID/experiment-runs" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg eid "$EXPERIMENT_ID" '{
"experiment_ids": [$eid],
"page_size": 20,
"selects": ["ID", "NAME", "STATUS", "INPUTS_PREVIEW", "OUTPUTS_PREVIEW"]
}')"
Page through results
Both examples below fetch up to 100 results across as many pages as that takes, then stop—so the two are comparable operations, not “one page” vs. “everything.” Adjust the100/page_size values for your own use case.
- Python
- TypeScript
- Java
- Go
- cURL
get_experiment_results paginates internally and stops once limit total results are returned. datasets.experiment_runs.query has no total-count limit; iterate the returned page with async for and break once you have enough.- Before
- After
Before
from langsmith import Client
client = Client()
experiment_id = client.read_project(project_name=experiment_name).id
# get_experiment_results paginated internally; increase `limit` to fetch
# more results in a single call. There is no cursor to pass in manually.
results = client.get_experiment_results(
project_id=experiment_id,
limit=100,
)
examples_with_runs = list(results["examples_with_runs"])
After
from langsmith import Client
import asyncio
async def main():
client = Client()
experiment_id = client.read_project(project_name=experiment_name).id
page = await client.datasets.experiment_runs.query(
str(dataset_id),
experiment_ids=[str(experiment_id)],
page_size=1,
)
runs = []
async for run in page:
runs.append(run)
if len(runs) >= 100:
break
return runs
examples_with_runs = asyncio.run(main())
The legacy dataset runs endpoint wasn’t exposed on the public TypeScript
Client. client.datasets.experimentRuns.query(...) returns an async iterable—use for await...of (no extra await needed) and break once you have enough.- Before
- After
Before
// The legacy dataset runs endpoint was not exposed on the public TypeScript Client.
// Use the cURL example for the old request body shape.
After
import { Client } from "langsmith";
const client = new Client();
const experimentId = (await client.readProject({ projectName: experimentName })).id;
const runs: unknown[] = [];
for await (const run of client.datasets.experimentRuns.query(datasetId, {
experiment_ids: [experimentId],
page_size: 1,
})) {
runs.push(run);
if (runs.length >= 100) break;
}
The legacy endpoint returns one page per call with no auto-pager—loop manually, incrementing
offset, and stop once you have enough. .experimentRuns().query(...).autoPager() walks pages for you—break out of the loop once you have enough runs.- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.datasets.runs.RunQueryParams
import com.langchain.smith.models.datasets.runs.ExampleWithRunsCh
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val examplesWithRuns = mutableListOf<ExampleWithRunsCh>()
var offset = 0L
val limit = 20L
while (true) {
val page = client.datasets().runs().query(
datasetId,
RunQueryParams.builder()
.addSessionId(experimentId)
.limit(limit)
.offset(offset)
.build()
).orElse(emptyList())
examplesWithRuns.addAll(page)
if (examplesWithRuns.size >= 100 || page.size.toLong() < limit) break
offset += limit
}
After
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.datasets.experimentruns.ExperimentRunQueryParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val page = client.datasets().experimentRuns().query(
datasetId,
ExperimentRunQueryParams.builder()
.addExperimentId(experimentId)
.pageSize(1L)
.build()
)
var count = 0
for (run in page.autoPager()) {
count++
if (count >= 100) break
}
The legacy endpoint returns one page per call with no auto-pager—loop manually, incrementing
Offset, and stop once you have enough. On the new endpoint, paginate manually by setting Cursor on the request from the previous response’s NextCursor and stopping once you have enough; avoid QueryAutoPaging here—it sends the cursor as a query parameter, which this POST endpoint doesn’t read, so it silently refetches the first page forever.- Before
- After
Before
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
var examplesWithRuns []langsmith.ExampleWithRunsCh
offset := int64(0)
limit := int64(20)
for {
page, err := client.Datasets.Runs.Query(ctx, datasetID, langsmith.DatasetRunQueryParams{
SessionIDs: langsmith.F([]string{experimentID}),
Limit: langsmith.F(limit),
Offset: langsmith.F(offset),
})
examplesWithRuns = append(examplesWithRuns, *page...)
if len(examplesWithRuns) >= 100 || int64(len(*page)) < limit {
break
}
offset += limit
}
After
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
params := langsmith.DatasetExperimentRunQueryParams{
ExperimentIDs: langsmith.F([]string{experimentID}),
PageSize: langsmith.F(int64(1)),
}
var examplesWithRuns []langsmith.DatasetExperimentRunQueryResponse
for {
page, err := client.Datasets.ExperimentRuns.Query(ctx, datasetID, params)
examplesWithRuns = append(examplesWithRuns, page.Items...)
if page.NextCursor == "" || len(examplesWithRuns) >= 100 {
break
}
params.Cursor = langsmith.F(page.NextCursor)
}
Raw HTTP has no auto-pagination helper: pass the previous response’s
next_cursor back in as cursor to fetch the next page.- Before
- After
Before
curl -X POST "https://api.smith.langchain.com/api/v1/datasets/$DATASET_ID/runs" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg eid "$EXPERIMENT_ID" '{
"session_ids": [$eid],
"limit": 20,
"offset": 20
}')"
After
curl -X POST "https://api.smith.langchain.com/v2/datasets/$DATASET_ID/experiment-runs" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg eid "$EXPERIMENT_ID" --arg cursor "$NEXT_CURSOR" '{
"experiment_ids": [$eid],
"page_size": 20,
"cursor": $cursor
}')"
Sort by feedback score
Sort dataset examples by a feedback score, supported only when you query a single experiment. In Go and Java, this replaces the legacysort_params.sort_by/sort_params.sort_order (now sort.by/sort.order); Python and TypeScript gain sorting for the first time in the new API.
- Python
- TypeScript
- Java
- Go
- cURL
get_experiment_results did not support sorting by feedback score.- Before
- After
# get_experiment_results did not support sorting results by feedback score.
After
from langsmith import Client
import asyncio
async def main():
client = Client()
experiment_id = client.read_project(project_name=experiment_name).id
page = await client.datasets.experiment_runs.query(
str(dataset_id),
experiment_ids=[str(experiment_id)],
sort={"by": "feedback.correctness", "order": "ASC"},
)
return page.items
examples_with_runs = asyncio.run(main())
The legacy dataset runs endpoint was not exposed on the public TypeScript
Client, so there was no way to sort by feedback score before the new API.- Before
- After
Before
// The legacy dataset runs endpoint was not exposed on the public TypeScript Client.
// Use the cURL example for the old request body shape.
After
import { Client } from "langsmith";
const client = new Client();
const experimentId = (await client.readProject({ projectName: experimentName })).id;
const page = await client.datasets.experimentRuns.query(datasetId, {
experiment_ids: [experimentId],
sort: { by: "feedback.correctness", order: "ASC" },
});
sortParams() is replaced by sort(), with sortBy()/sortOrder() renamed to by()/order().- Before
- After
Before
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.datasets.runs.RunQueryParams
import com.langchain.smith.models.datasets.runs.SortParamsForRunsComparisonView
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val examplesWithRuns = client.datasets().runs().query(
datasetId,
RunQueryParams.builder()
.addSessionId(experimentId)
.sortParams(
SortParamsForRunsComparisonView.builder()
.sortBy("correctness")
.sortOrder(SortParamsForRunsComparisonView.SortOrder.ASC)
.build()
)
.build()
)
After
import com.langchain.smith.client.LangsmithClient
import com.langchain.smith.client.okhttp.LangsmithOkHttpClient
import com.langchain.smith.models.datasets.experimentruns.ExperimentRunQueryParams
val client: LangsmithClient = LangsmithOkHttpClient.fromEnv()
val page = client.datasets().experimentRuns().query(
datasetId,
ExperimentRunQueryParams.builder()
.addExperimentId(experimentId)
.sort(
ExperimentRunQueryParams.Sort.builder()
.by("feedback.correctness")
.order("ASC")
.build()
)
.build()
)
SortParams is replaced by Sort, with SortBy/SortOrder renamed to By/Order.- Before
- After
Before
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
examplesWithRuns, err := client.Datasets.Runs.Query(ctx, datasetID, langsmith.DatasetRunQueryParams{
SessionIDs: langsmith.F([]string{experimentID}),
SortParams: langsmith.F(langsmith.SortParamsForRunsComparisonView{
SortBy: langsmith.F("correctness"),
SortOrder: langsmith.F(langsmith.SortParamsForRunsComparisonViewSortOrderAsc),
}),
})
After
package main
import (
"context"
"github.com/langchain-ai/langsmith-go"
)
ctx := context.Background()
client := langsmith.NewClient()
page, err := client.Datasets.ExperimentRuns.Query(ctx, datasetID, langsmith.DatasetExperimentRunQueryParams{
ExperimentIDs: langsmith.F([]string{experimentID}),
Sort: langsmith.F(langsmith.DatasetExperimentRunQueryParamsSort{
By: langsmith.F("feedback.correctness"),
Order: langsmith.F("ASC"),
}),
})
- Before
- After
Before
curl -X POST "https://api.smith.langchain.com/api/v1/datasets/$DATASET_ID/runs" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg eid "$EXPERIMENT_ID" '{
"session_ids": [$eid],
"sort_params": {
"sort_by": "correctness",
"sort_order": "ASC"
}
}')"
After
curl -X POST "https://api.smith.langchain.com/v2/datasets/$DATASET_ID/experiment-runs" \
-H "x-api-key: $LANGSMITH_API_KEY" \
-H "Content-Type: application/json" \
-d "$(jq -n --arg eid "$EXPERIMENT_ID" '{
"experiment_ids": [$eid],
"sort": {
"by": "feedback.correctness",
"order": "ASC"
}
}')"
Connect these docs to Claude, VSCode, and more via MCP for real-time answers.