Referencia
Límites y uso
Esta referencia está en inglés. La navegación y las etiquetas están traducidas.
Limits apply per key and count REST requests and MCP tool calls together. The data changes once a day, so caching goes a long way.
Limits
- Per key, per minute60 requests
- Per key, per day (UTC)5,000 requests
- Signed-in web session, per minute300 requests
- Public endpoints, per IP, per minute120 requests
Every response carries RateLimit-Limit, RateLimit-Remaining and RateLimit-Reset (seconds until the window resets) for the window closest to running out. Over the limit you get 429 rate_limited with Retry-After.
Need more? Write to us with what you are building.
Caching
- Responses for the same screen don't change. Cache them until
as_ofmoves;GET /screens/latestis a cheap way to check. - Every
GETreturns anETag. Send it inIf-None-Matchand an unchanged response comes back as304with no body. - Ask for what you need:
fieldstrims rows, andid in (…)fetches a known set in one call instead of one call per instrument.
Usage
Settings → API & MCP shows requests and MCP calls per day for the last 14 days, across your active keys, next to today's limit.
Usage is counted per key per UTC day. Revoking a key removes it from the chart.