Skip to content

docs: update API rate limits - #29

Open
samrathreddy wants to merge 1 commit into
fastino-ai:mainfrom
samrathreddy:docs/update-rate-limits
Open

docs: update API rate limits#29
samrathreddy wants to merge 1 commit into
fastino-ai:mainfrom
samrathreddy:docs/update-rate-limits

Conversation

@samrathreddy

Copy link
Copy Markdown
Member

Summary

  • update the public rate-limits page to match the current backend request-rate defaults
  • distinguish request-rate 429 responses from credit/spend-cap 402 responses
  • update synthetic data, training job, and API key endpoint pages so the inline limit notes match the consolidated table

Validation

  • git diff --check
  • stale-value search for old documented limits
  • mint dev --no-open served the updated rate-limits page at http://localhost:3001/api-reference/rate-limits
  • mint validate runs under Node 22, but fails on existing site config issues: favicon generation treats the remote DuckDuckGo favicon as a local path, and strict validation fails on an existing OpenAPI warning from docs.json.

@samrathreddy
samrathreddy marked this pull request as ready for review July 14, 2026 19:22
@jaanavit

jaanavit commented Aug 6, 2026

Copy link
Copy Markdown
Collaborator

Thanks for surfacing this — two of the facts here are real and currently missing from main, verified directly against the Brain source:

  1. POST /create-api-key is rate-limited to 50/hour — confirmed at brain/api/api_keys_router.py:103, @limiter.limit("50/hour", key_func=generation_user_key). Not documented anywhere on the site today.
  2. POST /generate has a second, global limit on top of the per-user one500/min shared across all callers platform-wide, not just your own 120/min budget. Confirmed at brain/api/felix_generate_router.py:90-95, @limiter.shared_limit(generation_global_limit, scope=GENERATION_GLOBAL_SCOPE, ...). api-reference/synthetic-data.mdx and guides/synthetic-data.mdx still only mention the 120/min per-user figure with no hint of the shared ceiling.

The rest of this PR's rate-limits.mdx changes are based on an older state of the page (the 1,000/min defaults, the daily_spend_cap_exceeded header framing) that's since been superseded by #47 — main now has the corrected numbers plus the actual 402/403 credit-cap mechanism, so merging this as-is would regress that. Given the conflicts GitHub is already flagging, we're going to close this rather than rebase it, but I'll open a follow-up that folds in the two real findings above (the create-api-key limit + the generate global cap) on top of the current page. Appreciate the catch on those two.

@jaanavit jaanavit closed this Aug 6, 2026
@jaanavit jaanavit reopened this Aug 6, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants