| TIME | STATUS | MODEL | PROMPT | OUTPUT | TOK/S | DURATION | FINISH |
|---|---|---|---|---|---|---|---|
| connecting… | |||||||
| MODEL | REQUESTS | ERRORS | PROMPT | OUTPUT |
|---|---|---|---|---|
| no requests yet | ||||
/v1/chat/completions on this host. Nothing is stored — the result shows here only.response appears here…
Thin OpenAI-compatible proxy for the opencode.ai free tier. Runs locally, exposed through a Cloudflare Quick Tunnel. Zero dependencies.
GET /v1/models — free models, fetched live from upstream (5 min cache; requires the API key)POST /v1/chat/completions — OpenAI shape: messages, tools, tool_choice, streamGET /api/health — deployment checkGET /api/stats — in-memory request stats for the dashboard (resets on cold start, nothing stored)Source: github.com/MarkDev11/oc-proxy. Long requests close gracefully with finish_reason: "length" so agent loops can continue.