Skip to main content
Stuck on a specific error or missing data? Start here.
Your X-Toolken-Key is missing or invalid. Check that the header is present on every request and that the key value is correct.
Your X-Toolken-Metadata contains an invalid key or exceeds the size or key-count limit. Keys must be lowercase, start with a letter, and use snake_case. Note: malformed JSON is silently ignored, so verify your JSON is well-formed before checking key names.
You hit a request rate limit or the concurrent-stream limit. Read the Retry-After header and back off before retrying.
Confirm your base URL ends in /v1, the X-Toolken-Key header is present, and the model name routes to a recognized provider. If the model name is ambiguous, set the X-Toolken-Provider header explicitly.
Use uri_base: "https://gateway.toolken.ai/v1/" (with the trailing slash) and pass Toolken headers via extra_headers. The base_url and default_headers options are not honored by ruby-openai.
Point a raw net/http request at https://gateway.toolken.ai/v1/chat/completions with the headers set directly, or verify that your configured base URL preserves the /v1/ path segment.
The Anthropic SDK base URL is https://gateway.toolken.ai with no /v1. The SDK appends /v1/messages itself, so adding /v1 to the base URL produces a doubled path and fails.
For the full list of error codes, see the Errors reference. Still stuck? Email [email protected].