FuelAPI reference
Every operation the guides use, in one place. Plain HTTPS and JSON.
Management API
Base https://fuel.domain/api/project/v1, with Authorization: Bearer fl_sk_.... Every POST takes an Idempotency-Key.
| Operation | Does |
|---|---|
POST /launches/prepare |
Prepare a launch for the creator's wallet to sign |
POST /launches/{id}/submit |
Send a signed launch or import |
GET /launches/{id} |
Read a launch and its state |
POST /tokens/import |
Prepare bringing an existing pump.fun token |
POST /presets |
Make a fee strategy |
GET /accounts/{id} |
Read a token's credit account |
GET /accounts/{id}/ledger |
Every change to an account |
POST /accounts/{id}/purchases |
Buy credit in USDC |
POST /keys |
Make a router key |
PATCH /keys/{id} |
Rename a key or change its limit |
DELETE /keys/{id} |
Revoke a router key |
GET /usage |
Router usage by day, account, key and model |
GET /statements/{month} |
A month's statement |
POST /webhooks |
Add a webhook endpoint |
POST /webhooks/{id}/test |
Send a ping |
POST /webhooks/{id}/rotate |
Rotate the signing secret |
GET /webhooks/deliveries |
List deliveries |
POST /webhooks/deliveries/{id}/replay |
Send a delivery again |
Fuel Router
Base https://fuel.domain/api/v1, with Authorization: Bearer fl_ai_... or x-api-key.
| Operation | Does |
|---|---|
POST /chat/completions |
OpenAI Chat Completions, every model |
POST /responses |
OpenAI Responses, stateless |
POST /messages |
Anthropic Messages, every model |
POST /images |
Make images |
GET /models |
Every model with prices, no key needed |
GET /images/models |
Image models, their fields and prices |
GET /balance |
The key's account balance |
Errors
| Status and code | Means |
|---|---|
402 insufficient_balance |
The account can't cover the call's reserve |
402 key_limit |
The key's spending limit can't cover the reserve |
400 model_not_served |
Free, batch or retired versions, and after-the-fact routers |
400 unknown_model |
The router doesn't know this model |
400 tool_not_served |
Hosted tools that keep state or call other models |
400 image_type_not_served |
Inline image types other than PNG, JPEG, WebP, GIF |
400 model_no_image_input |
The model can't read images |
400 import_refused |
The token can't be brought |
503 model_busy / 529 |
Every provider is busy; wait Retry-After-Ms |
Rate limits
An account runs at most 8 calls at once and starts at most 600 a minute. A refused call is free, and its answer names the wait.