OmniaKey returns protocol-compatible errors where possible, so your existing SDK error handling continues to work.Documentation Index
Fetch the complete documentation index at: https://docs.omniakey.com/llms.txt
Use this file to discover all available pages before exploring further.
Common status codes
| Status | Meaning |
|---|---|
400 | Invalid request body, unsupported field, or malformed parameter |
401 | Missing or invalid API key |
403 | Account balance, API key cap, model access, or policy restriction blocks the request |
404 | Wrong path, wrong base URL suffix, or unknown model id |
429 | OmniaKey or upstream provider throttling |
500 | OmniaKey or upstream provider error |
Base URL mistakes
Most setup errors come from using the wrong protocol suffix:| Tool or SDK | Correct base URL |
|---|---|
| OpenAI-compatible | https://api.omniakey.com/v1 |
| Anthropic / Claude Code | https://api.omniakey.com |
| Gemini native | https://api.omniakey.com/v1beta |
/v1/messages, the base URL should be the bare
host. If an OpenAI SDK gets a 404, confirm it includes /v1.
Quota errors
A403 can mean either:
- the account balance is too low, or
- the API key’s per-key cap has been reached.
Rate limits
OmniaKey supports route-level and model-request throttling, and upstream providers can also return429 through the relay. Public docs do not publish
fixed thresholds because limits can vary by route, account/group configuration,
and provider.
Treat 429 as a retryable throttling signal: retry with exponential backoff,
add jitter, and avoid sending the same burst again immediately.