Joy ships with exactly one compiled-in inference provider: the Joy relay (bentoo), an OpenAI-compatible endpoint at https://api.bentoo.ai/v1. It serves three Qwen model tiers — fast, thinking, and pro — so a fresh install needs no third-party accounts, only a relay API key.

​
The Built-in Relay

The relay’s base URL is fixed in the compiled-in catalog; there is nothing to configure beyond the key. Authentication uses a single API key read from the JOY_RELAY_API_KEY environment variable:
Never paste the key itself into a config file — Joy only stores the name of the environment variable holding it, and raw keys are rejected at parse time.

​
Configuring the Relay

During onboarding the provider menu offers the built-in relay — pick a tier and paste your key. You can also configure it from the command line:
Or from Settings → Inference in the web interface. Whichever route you take, the selection is written to config.toml as a model slot:
api_key_env names the environment variable holding your key. If no slot is configured, Joy falls back to the relay defaults — the Thinking tier — and reads the key from JOY_RELAY_API_KEY.

​
Choosing Your Model

When an administrator has configured a model allowlist, every user can choose their own model under Settings → Inference. The preference applies to that user’s future messages in every conversation; it does not change the provider or another user’s preference. Administrators configure the allowlist from Settings → Inference → User model access. Add model IDs manually or fetch the active provider’s model catalog, select the models users may choose, set the workspace default, and save. Provider configuration alone does not publish models to users; this explicit step keeps newly available provider models outside the tenant allowlist until an administrator approves them. The same preference is available from chat commands:
/model shows the current preference, effective model, and allowed models. /model default returns to the workspace default. Only models from the active provider’s administrator-approved allowlist can be selected.

​
Adding Providers

The built-in relay is the only provider Joy ships with, and it is all most installs need. If you must reach a different backend — a self-hosted vLLM endpoint, an internal gateway, or another vendor — register it as a catalog entry in providers.json; entries there are appended to the compiled-in catalog and later entries override earlier ones by id. See Configuration for the file’s location and precedence.