One endpoint.
Every free model.
Relay pools your free providers behind a single OpenAI-compatible API — automatic failover, per-request routing, and live health. Running on the edge, even with your PC off.
| BASE | | |
|---|---|---|
| CHAT | | |
| IMAGES | | |
| AUDIO | | |
| VIDEO | | |
| EMBED | |
Providers
Status dots come from the gateway's own probe history. Select a provider for details.
Gateway models
Model ids are provider/model. auto fails over across providers.
Health: working · unstable · down.
Select a provider dot for probe history. Each row has explicit enable/favorite/copy/play actions.
Admin
Manage provider channels, keys, usage and configuration for this gateway.
Admin token
Stored in this browser only. With a valid token your model visibility prefs sync to the gateway.
Ask the operator for the token, or set your own ADMIN_TOKEN secret on the Worker.
Channels
Each channel = one provider endpoint + any number of API keys. On error the gateway rotates to the next key, then the next channel.
+ Add channel
Usage & logs
Counters are isolate-local — a lower bound that can reset any time. Request logs are capped per isolate.
Config backup
Export or restore your channel configuration as JSON.