Providers

The kinds of provider in Add provider…, in Settings. Each bot runs on one provider.

Kind	Connects with	Base URL	Models from
ChatGPT plan	Device code		The Codex model list
Grok plan	Device code		Grok's model list
Vercel account	Device code, then a team. Needs `FXBOT_VERCEL_CLIENT_ID`.		AI Gateway
AI Gateway key	Key		AI Gateway
OpenAI key	Key	`https://api.openai.com/v1`	models.dev
OpenCode Zen key	Key	`https://opencode.ai/inference/openai/v1`	models.dev, Chat Completions models only
OpenCode Go key	Key	`https://opencode.ai/inference/go/openai/v1`	models.dev, Chat Completions models only
OpenRouter key	Key	`https://openrouter.ai/api/v1`	models.dev
Ollama	Base URL, and an optional context window	`http://localhost:11434/v1` by default	The server: `/api/tags`, and `/api/show` for each model
LM Studio	Base URL, and an optional context window	`http://localhost:1234/v1` by default	The server: `/api/v1/models`
Other server	Base URL, and an optional key and context window		The server: `/models`, ids only

Keys and sign-ins

A key is checked with the provider when it is added, at its model list. It is refused when the provider answers 401 or 403, and kept on a 404, which a server with no list at /models gives.
A key goes to the server once and is never sent back. A provider in the list says only whether it holds a key or a sign-in.
A device code sign-in runs on the server. The page gets the link, the code, and when it expires.
Keys and tokens are stored sealed. See the security model.

fxbot keeps each provider's model list in Postgres and answers with it at once. A list older than 5 minutes is fetched again behind that answer, and a failed fetch keeps the old list.
A list is dropped when its source changes: a new key, address, team, or sign-in.
models.dev is read at most once a day.
The model picker takes a model id that the list lacks.
A provider's Context window, on Ollama, LM Studio, and Other server, overrides what the server reports, for every model it serves.