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: Tensorlake
providerId: tensorlake
created: '2026-07-12'
modified: '2026-07-12'
reconciled: false
tags:
- Document Extraction
- Document Ingestion
- OCR
- AI
- Rate Limiting
- Quotas
description: >-
  Tensorlake does not publish fixed numeric per-minute request-rate limits for
  the Document Ingestion API. In practice throughput is governed by plan-level
  concurrency (for example up to 2 concurrent sandboxes on Free, up to 100 on
  On-Demand, and up to 1,000 on Pro) and by the credit / page allowance of the
  plan rather than by a per-endpoint request cap. Parsing and extraction run as
  asynchronous jobs, so large workloads are absorbed by the job queue and read
  back by polling or webhook rather than by holding open synchronous requests.
notes: >-
  No explicit numeric API rate limits are documented as of the review date.
  Concurrency ceilings are drawn from the pricing page and should be verified.
  Assume standard 429 throttling with backoff on sustained bursts.
sources:
- https://www.tensorlake.ai/pricing
- https://docs.tensorlake.ai/api-reference/introduction
- https://docs.tensorlake.ai/platform/authentication
responseCodes:
  throttled: 429
limits:
- name: API Request Rate
  scope: account
  metric: requests
  limit: not published
  notes: No fixed numeric request-rate limit is documented for the REST API.
- name: Concurrent Sandboxes
  scope: account
  metric: sandboxes
  limit: per plan (Free ~2, On-Demand ~100, Pro ~1,000)
  notes: Concurrency ceiling is set by the subscription tier.
- name: Document Page Allowance
  scope: account
  metric: pages
  limit: per plan / credit balance
  notes: Document ingestion is metered by the page against the plan's credit allowance.
policies:
- name: Asynchronous Jobs
  description: Parse and extraction requests return a job id immediately; results are retrieved by polling GET /documents/v2/parse/{parse_id} or via webhook, smoothing bursty load.
- name: Backoff Strategy
  description: Clients should implement exponential backoff with jitter and honor Retry-After on 429 responses.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com