specification: API Commons Rate Limits
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits
provider: Azure Databricks
providerId: azure-databricks
created: '2026-05-04'
# Provenance stamped 2026-08-11: this artifact was written by the API Evangelist
# bulk sweep dated 2026-05-04, not harvested from the provider. See roadmap#35.
method: generated
modified: '2026-05-05'
reconciled: true
tags:
  - Analytics
  - Big Data
  - Lakehouse
  - Spark
  - Rate Limiting
description: Azure Databricks publishes detailed per-API rate limits at the workspace and account level
  for Jobs, Workspace, MLflow, SCIM, and other endpoints. Most limits are fixed; resource limits (clusters,
  jobs, MLflow runs, Vector Search) are also enforced. Throttled requests return HTTP 429. Some limits
  are negotiable through your Databricks account team.
sources:
  - https://learn.microsoft.com/en-us/azure/databricks/resources/limits
headers:
  retryAfter: Retry-After
responseCodes:
  throttled: 429
limits:
  - name: DBFS API
    scope: workspace
    metric: requests_per_second
    limit: 30
    timeFrame: second
  - name: Jobs API - create
    scope: workspace
    metric: requests_per_second
    limit: 20
    timeFrame: second
  - name: Jobs API - run-now
    scope: workspace
    metric: requests_per_second
    limit: 20
    timeFrame: second
  - name: Jobs API - runs/get
    scope: workspace
    metric: requests_per_second
    limit: 100
    timeFrame: second
  - name: Jobs API - runs/submit
    scope: workspace
    metric: requests_per_second
    limit: 35
    timeFrame: second
  - name: Account SCIM API - GET
    scope: account
    metric: requests_per_second
    limit: 20
    timeFrame: second
  - name: Account SCIM API - LIST
    scope: account
    metric: requests_per_minute
    limit: 240
    timeFrame: minute
  - name: Account SCIM API - PATCH
    scope: account
    metric: requests_per_second
    limit: 2
    timeFrame: second
  - name: Account SCIM API - POST/PUT/DELETE
    scope: account
    metric: requests_per_second
    limit: 5
    timeFrame: second
  - name: Workspace SCIM API - GET
    scope: workspace
    metric: requests_per_minute
    limit: 255
    timeFrame: minute
  - name: Permissions API - GET
    scope: workspace
    metric: requests_per_second
    limit: 100
    timeFrame: second
  - name: Permissions API - PATCH/PUT
    scope: workspace
    metric: requests_per_second
    limit: 30
    timeFrame: second
  - name: Secrets API
    scope: workspace
    metric: requests_per_minute
    limit: 1100
    timeFrame: minute
  - name: Token Management API
    scope: workspace
    metric: requests_per_second
    limit: 40
    timeFrame: second
  - name: MLflow Tracking API (most endpoints)
    scope: workspace
    metric: requests_per_second
    limit: 120
    timeFrame: second
  - name: MLflow search endpoints
    scope: workspace
    metric: requests_per_second
    limit: 7
    timeFrame: second
  - name: Workspace export API
    scope: workspace
    metric: requests_per_second
    limit: 60
    timeFrame: second
  - name: Workspace import API
    scope: workspace
    metric: requests_per_second
    limit: 30
    timeFrame: second
  - name: Genie API (free tier)
    scope: workspace
    metric: questions_per_minute
    limit: 5
    timeFrame: minute
    notes: Best-effort across all Genie spaces; raise via Databricks account team.
  - name: Genie UI questions
    scope: workspace
    metric: questions_per_minute
    limit: 20
    timeFrame: minute
  - name: MCP servers (managed UC/VectorSearch/Genie)
    scope: workspace
    metric: queries_per_second
    limit: 50
    timeFrame: second
  - name: MLflow Trace creation
    scope: workspace
    metric: queries_per_second
    limit: 50
    timeFrame: second
  - name: Model Serving throughput target
    scope: endpoint
    metric: queries_per_second
    limit: 200
    timeFrame: second
    notes: Soft target; not guaranteed.
policies:
  - name: Backoff
    description: Honor Retry-After on 429 responses; use exponential backoff with jitter.
  - name: Limit raise via account team
    description: Many limits marked "Fixed=No" can be raised by contacting your Azure Databricks account
      team.
  - name: Workspace vs account scope
    description: Some limits are workspace-scoped, others account-scoped; check the per-API table.
  - name: Resource limits separate from rate limits
    description: Azure Databricks also enforces resource caps (clusters, MLflow runs, Vector Search indexes,
      Unity Catalog objects) that are not API rate limits but block create operations once exceeded.
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com