name: Benchling Rate Limits
description: Benchling enforces both request-rate limits (requests per 30 seconds) and throughput limits (objects created or updated per hour) to ensure platform stability. Limits vary by key type. Apps have independent per-app limits so they do not consume a shared tenant budget.
url: https://docs.benchling.com/docs/rate-limiting

rate_limits:

  - name: User API Key Limit
    description: Aggregate limit across all personal/user API keys within a single tenant.
    limit: 60
    window: 30 seconds
    scope: per-tenant (all user keys combined)
    key_type: user_api_key

  - name: App API Key Limit
    description: Per-app limit for individual Benchling App keys. Apps have independent budgets so one heavy app does not throttle others.
    limit: 300
    window: 30 seconds
    scope: per-app-key
    key_type: app_client_credentials

  - name: Combined App Tenant Limit
    description: Aggregate ceiling for all Benchling Apps operating within a single tenant. This limit is not reflected in response headers.
    limit: 1000
    window: 30 seconds
    scope: per-tenant (all apps combined)
    key_type: app_client_credentials
    note: Not exposed via rate-limit response headers.

  - name: Platform-Wide Throughput Limit
    description: Limits the total number of objects (entities, containers, results, etc.) created or updated across all API calls over a rolling hour window. Applies to entire tenant and adjusts dynamically under system load.
    limit: tens of thousands of objects
    window: 1 hour
    scope: per-tenant (all keys combined)
    dynamic: true
    note: Exact threshold may change in response to system load. Most tenants operate well below this ceiling.

response_headers:
  - name: x-rate-limit-limit
    description: Total requests allowed in the current window period.
  - name: x-rate-limit-remaining
    description: Remaining requests available in the current window.
  - name: x-rate-limit-reset
    description: Seconds until the current rate-limit window resets.

error_handling:
  http_status: 429
  error_type: invalid_request_error
  message: "Rate limit exceeded."
  recommended_strategy: Exponential backoff with randomized jitter; maximum recommended delay of ~15 seconds.

escalation:
  description: Rate limits can be increased upon request to Benchling support for tenants with legitimate high-volume use cases.
  contact: https://benchling.com/contact