Skip to main content

​
Credits

The API uses the credits of the connected Havincy workspace, at the same price as in the app.
  • Each model has a cost per output: credits_per_output in GET /models.
  • POST /generations debits the credits when the job is created and returns credits_charged and the new credits_balance.
  • If the generation fails on the provider side, the credits are refunded automatically.
  • Some monthly gifts (“Sparks”) can cover a generation: the response then contains covered_by_monthly_gift: true and credits_charged: 0.
  • When the balance is too low, the API returns 402 insufficient_credits with the required amount in details.
Check the balance at any time with GET /account. Credits are bought or renewed through a Havincy subscription or credit pack.

​
Rate limits

Requests are limited per workspace, per minute, across the REST API and the MCP server combined. The limit depends on the workspace plan: Every response includes:
When the limit is reached, the API returns 429 rate_limited with a Retry-After header (seconds). MCP protocol messages (initialize, tools/list, ping) do not count.

​
Other limits

Need higher limits? Contact support@havincy.com.