generated: '2026-09-19'
method: searched
source: https://nefesh.ai/docs/api
docs:
- https://nefesh.ai/docs/api
- https://nefesh.ai/terms
- https://nefesh.ai/pricing
- https://nefesh.ai/docs/v1-ingest
limit_count: 5
limits:
- scope: per-session
  window: 1 minute
  limit: 120
  unit: requests
  tier: Solo (and the terms' default)
  source: 'https://nefesh.ai/docs/api ("Per session 120 req/min"); terms §5 "120 requests per minute per session"'
- scope: per-key
  window: 1 minute
  limit: 10
  unit: requests
  tier: Free
  source: 'https://nefesh.ai/pricing ("10 req/min"); MCP request_api_key tool description'
- scope: per-key
  window: 1 minute
  limit: 50
  unit: new sessions
  tier: all
  source: 'https://nefesh.ai/docs/api ("New sessions per key 50/min"); terms §5 "50 concurrent sessions per minute per API key"'
- scope: per-account
  window: 1 month (billing period)
  limit: 1000 (Free) / 50000 (Solo) / unlimited (Enterprise)
  unit: API calls
  source: https://nefesh.ai/pricing
- scope: per-request
  window: n/a
  limit: 4 KB
  unit: request body
  source: 'https://nefesh.ai/docs/v1-ingest ("Maximum 4 KB per request")'
responses:
  rate_limit_exceeded:
    status: 429
    body: '{"detail": "Rate limit exceeded"} (docs common-errors table: "429: Rate limit exceeded — More than 120 req/min — Throttle requests")'
  quota_exceeded:
    status: 402
    body: 'terms §5: "Exceeding your monthly quota will result in HTTP 402 responses until additional usage is purchased or the next billing period begins."'
  body_too_large:
    status: 413
headers:
  documented: []
  observed: []
  note: >-
    No RateLimit-*, X-RateLimit-* or Retry-After header is documented anywhere, and none appeared on the live 401
    responses probed on 2026-09-19 (only Cloudflare cf-ray / cf-cache-status and security headers). An agent has no
    runtime budget signal before the 429 — it must count calls itself against the documented numbers.
enterprise:
  note: 'Rate limits are "Custom (unlimited available)" on the Enterprise tier; Nefesh also reserves the right to throttle or suspend access on abuse patterns (terms §5).'