Tarseek API

Errors

Every public chat error uses one stable envelope:

{
  "error": {
    "message": "Human-readable explanation",
    "type": "authentication_error",
    "code": "invalid_api_key",
    "param": null,
    "retryable": false
  },
  "request_id": "req_..."
}

x-request-id matches the body request_id. Send an optional printable-ASCII X-Client-Request-Id (maximum 512 characters) to correlate the call with your own logs. Retryable provider throttling includes Retry-After and error.retry_after_seconds when a delay is known.

HTTP	Code	Type	Retry	Cause	Action
400	`validation_error`	`invalid_request_error`	No	Invalid JSON or request schema	Fix the request
401	`invalid_api_key`	`authentication_error`	No	Missing, unknown, or disabled key	Use an active project key
402	`insufficient_credits`	`billing_error`	No	Prepaid balance too low	Add credits
404	`model_not_found`	`invalid_request_error`	No	Unknown or unlisted model ID	Use the catalog
400	`unsupported_parameter`	`invalid_request_error`	No	Parameter not supported by this model	Remove the field
400	`context_length_exceeded`	`invalid_request_error`	No	Prompt longer than context	Shorten input
429	`rate_limit_exceeded`	`rate_limit_error`	Yes	Project or key RPM/TPM/concurrency limit	Backoff using Retry-After
429	`provider_rate_limit`	`rate_limit_error`	Yes	Upstream 429 after failover exhausted	Retry later or another model
503	`model_unavailable`	`service_unavailable_error`	Yes	No healthy deployment	Retry or switch model
502	`provider_error`	`api_error`	Yes	Upstream failure after failover	Retry with backoff
502	`provider_authentication_error`	`provider_error`	Yes	Tarseek’s provider credential failed	Report to support; try another model
504	`provider_timeout`	`timeout_error`	Yes	Upstream deadline	Retry; reduce max output
500	`internal_error`	`internal_error`	Yes	Unexpected Tarseek error	Retry; contact support with request_id