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: Lightly
providerId: lightly-ai
created: '2026-06-21'
modified: '2026-06-21'
reconciled: false
tags:
- AI
- Computer Vision
- Data Curation
- Active Learning
- Embeddings
- Rate Limiting
- Quotas
- Throttling
description: >-
  Lightly does not publish explicit numeric rate limits for the LightlyOne
  platform REST API at https://api.lightly.ai. Practical limits are governed
  mainly by plan-level quotas - input dataset size (~25k samples on Community,
  ~100k on Team, unlimited on Enterprise) - and by concurrency of LightlyOne
  Worker compute runs rather than per-second request ceilings. Clients should
  treat the API as quota-bounded and handle HTTP 429 with backoff. Specific
  per-endpoint request limits are not reconciled in this artifact.
notes: >-
  Confirm any per-token request limits and worker concurrency caps with Lightly
  on reconciliation; the documented constraints are dataset-size quotas tied to
  plan tier, not published RPM/TPM values.
sources:
- https://www.lightly.ai/pricing
- https://docs.lightly.ai/
- https://docs.lightly.ai/lightly.api.html
responseCodes:
  throttled: 429
limits:
- name: Input Dataset Size (Community)
  scope: account
  metric: samples
  limit: 25000
  notes: Indicative free-tier dataset quota; verify on reconciliation.
- name: Input Dataset Size (Team)
  scope: account
  metric: samples
  limit: 100000
  notes: Indicative Team-tier dataset quota; verify on reconciliation.
- name: Input Dataset Size (Enterprise)
  scope: account
  metric: samples
  limit: unlimited
  notes: Enterprise removes dataset-size quotas; negotiated.
- name: Worker Run Concurrency
  scope: account
  metric: runs
  limit: see provider documentation
  notes: Concurrent LightlyOne Worker compute runs depend on registered workers and plan.
- name: Request Rate
  scope: account
  metric: requests
  limit: see provider documentation
  notes: No public numeric RPM/TPM published for the REST API; honor 429 / Retry-After.
policies:
- name: Quota-Based Limiting
  description: Effective limits are driven by plan-level dataset-size quotas and worker concurrency rather than published per-second request ceilings.
- name: Backoff Strategy
  description: Clients should implement exponential backoff with jitter and honor Retry-After on HTTP 429.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com