FAQ and Support
Find answers to common integration, billing, and account questions.
Is the API compatible with OpenAI?
Yes. Start with Chat Completions or Embeddings by changing the Base URL, API key, and model ID. The gateway also accepts supported Anthropic and Gemini client protocols.
How do I switch models?
Change the request's public model ID. Check the Models page because capabilities, limits, availability, and prices differ.
Why is part of my balance frozen?
Frozen credits are temporary reservations for requests still in progress. They are released or replaced by the final settled charge.
Are failed requests charged?
Requests with no billable usage are not charged. Output or usage already produced before an interruption can still be billed. The request log is the source of truth for the final amount.
Why can I see request and response bodies in the logs? Is my data safe?
To help you verify request parameters, inspect model outputs, and troubleshoot errors, sanitized request and response bodies are retained briefly (with binary, base64, and oversized content replaced by placeholders) and are permanently deleted automatically once expired. Your request and response content is kept strictly for troubleshooting, is never used to train any AI model or sold to third parties, and is not inspected by staff unless you authorize troubleshooting via a support ticket. See Data Privacy Policy for details.
What should I do if an API key leaks?
Disable or delete it immediately, create a replacement, update your service, and review recent request logs.
How do I contact support?
Submit a ticket from the feedback entry in the console. Include the request ID or task ID, time of the issue, model ID, observed behavior, and expected result. Never send a complete API key or password.
Help us improve this page
Found something unclear, outdated, or incorrect?
Last updated on