Zen Proxy — the free-tier hype machine

Requests served

–

0 errors

Upstream

–

Uptime

–

seconds online

Last minute

–

– in 5 min

Models

–

– allowed

Recent requestslive feed

Time	Model	Result	Latency	Count
no requests yet — go make some noise.

Allowed models

These are served on /v1/models and accepted by the proxy. Hit TEST to send a tiny completion and see live status. The default model is used when a request asks for an unknown one. Auto-sync probes every upstream free model and prunes the dead ones automatically.

auto-sync: …

Model	Default	Last test

Server settingsport / host need restart

Host Port Upstream url User-agent (the unlock)

Default model (empty = auto-pick healthy free model) Proxy key (empty = open) Zen key (byok, optional)

Timeout (ms)

Fallback models (json array — tried in order on 429/5xx) Model aliases (json map — e.g. {"gpt-4o":"space-bunny-free"})

Trust x-forwarded-for from a reverse proxy

Auto-sync models (probe upstream free models, prune dead ones)

Auto-sync interval (ms) Health-probe credentials

Auto-track opencode releases & update the User-Agent

User-agent refresh interval (ms)

Inject x-opencode-session header per client (free tier needs it)

Rate-limit chat requests per client (0 = off)

Requests / window

Window (ms)

Responses-API models (json array — patterns may end in *)

Server logs

Loading…

Locked 🔒

Recent requestslive feed

Allowed models

Server settingsport / host need restart

Server logsRefresh

Server logs