–
Total requests
–
Requests today
–
Errors
–
Tokens (in + out)
–
Tokens saved (Token Saver)
Usage by model
| Model | Requests | Prompt tokens | Completion tokens | Errors |
|---|
Usage by provider
| Provider | Requests | Errors | Avg latency |
|---|
Providers
Combos (fallback chains)
A combo name is what your tools use as the model. Steps are tried in order; on quota/rate-limit/server errors the next step takes over.
Request log
| Time | Requested | Provider | Upstream model | Status | Latency | Tokens in/out | Attempts |
|---|
Settings
Connect a tool
Point any OpenAI-compatible tool (Claude Code, Cursor, Cline, OpenCode…) at this base URL with the gateway key, and use a combo name as the model.