Rate limits
How many requests a key can make, and what happens when it goes over.
Today
Each API key can make about 60 requests a minute, across the REST API and the MCP server together.
When a key goes over, Hobson refuses its requests until the minute is up. At the moment that refusal comes back as 401 Unauthorized (and invalid_token on MCP), the same as a bad key. If a working key suddenly gets 401s, wait a minute before you replace it.
Per-plan limits
Coming soonLimits will scale with your plan and count per key and per workspace:
| Plan | Per key / minute | Workspace / minute | Calls at once | Searches / billing period | Exports / day |
|---|---|---|---|---|---|
| Starter | 120 | 240 | 4 | 3,000 | 2 |
| Growth | 300 | 600 | 8 | 10,000 | 5 |
| Scale | 600 | 1,200 | 16 | 30,000 | 10 |
| Pro | 1,200 | 2,400 | 32 | 100,000 | 20 |
Trials get Starter limits. Heavier calls, such as search, drafting and exports, will count as more than one request.
When this ships, going over a limit returns 429 Too Many Requests with a Retry-After header that says how many seconds to wait. Responses will also carry RateLimit headers so you can slow down before you hit the limit. On MCP, Claude gets a tool error that says when to try again.
Handling limits well
- Wait and retry rather than sending the same request again straight away. Back off a little longer each time.
- Page through lists instead of fetching one item at a time.
- Use one key per integration so one busy script can't use up another's allowance.