Skip to main content
Gateway calls are paid with the same pre.dev credits as coding, planning, and browser tasks. 1 credit = $0.10.

​
How a call is priced

Each call is charged in credits based on the model’s metered cost for that call. The metered cost is reported in dollars as usage.cost on every inference response, and the credits charged are reported in the x-predev-credits-charged header. The credit price per model is listed on GET /v1/models: each row carries a predev object with credits_per_m_input and credits_per_m_output, and, where the model has them, credits_per_m_cache_read, credits_per_m_cache_write, credits_per_image, and credits_per_request. Read those numbers rather than deriving prices yourself; they are what the charge is computed from. Worked example: a model whose row lists credits_per_m_input: 28.6 and credits_per_m_output: 142.9 costs 0.286 credits for 10,000 input tokens plus 0.286 credits for 2,000 output tokens, so a call of that size settles at about 0.57 credits.

​
What is included

  • Plugin costs, such as web search, are part of usage.cost and therefore of the charge.
  • With models fallbacks or provider preferences, you pay the price of the model and provider that actually served the call.
  • Streams settle at the final chunk. If your client disconnects mid-stream, the call is still charged for what the model generated.
  • Video jobs are charged when a status poll reports completion. See images and video.
  • Catalog reads (/v1/models, /v1/providers, and the per-modality model lists) are free and not rate-limited.
Each inference response reports the charge in x-predev-credits-charged and the balance in x-predev-credits-remaining. GET /v1/usage breaks the last 1–90 days down by model and path.

​
Trial

A free workspace can make 5 credits of gateway calls in total. After that, every call returns 402 with code subscription_required until the workspace subscribes. Trial workspaces are also limited to 30 requests per minute.

​
Out of credits

A workspace with no credits gets 402 with code insufficient_credits before anything is sent to the model, so nothing is generated and nothing is charged. The error body includes credits_remaining, estimated_credits, and a topup_url. See errors. Turn on auto-recharge in dashboard billing to keep a deployed app running when the balance would otherwise reach zero.

​
Rate limit

Requests are limited per workspace per minute: 600 by default, 30 on trial. A 429 carries a Retry-After header. Catalog reads do not count. See plans and credits for plan allowances.