Model Providers

Models are configured per Agent. Open Agents, select an Agent, then use Agent > Models to connect provider credentials, choose model roles, and test the route. The same connection flow is available from the CLI:

fased providers connect
fased providers connect openai

There is one provider registry shared by onboarding, CLI, Agent setup, Chat, tasks, and channel-routed Agents. Channels such as Telegram or Discord are separate from model providers; configure those from Agent > Channels.

Model Route

What This Section Covers

Page	Use it for
Model Provider Quickstart	First setup path from credential to first chat.
Models	Model selection, aliases, fallbacks, session overrides, and CLI commands.
Model providers	Provider registry rules, model metadata, refresh, and local endpoints.
Model failover	Auth profile rotation and fallback behavior.

Provider Groups

Group	Meaning
Common providers	Main hosted providers most users start with.
Local and private	Ollama, LM Studio, vLLM-compatible, LiteLLM, and Custom Provider routes.
Additional providers	Supported hosted or gateway providers that remain first-class but are less common first-run choices.

Code-Backed Provider Surface

These pages map to the shared provider registry used by onboarding, CLI, and Agent > Models. If a provider is not visible as ready, first check that the Agent has a credential, token, endpoint, or local server configured for that provider.

Provider page	Registry id	Main route or surface
OpenAI	`openai`	`openai`, `openai-codex`
Anthropic	`anthropic`	`anthropic` with OAuth, setup-token, or API key auth
Chutes	`chutes`	`chutes` with sign-in or API key
Ollama	`ollama`	native Ollama URL
LM Studio	`lmstudio`	local LM Studio `/v1` server
vLLM-compatible	`vllm`	OpenAI-compatible local server
MiniMax	`minimax`	`minimax`, `minimax-cn`, `minimax-portal`
Moonshot AI	`moonshot`	`moonshot`, `kimi-coding`
Google	`google`	`google`, `gemini`, `google-gemini-cli`
xAI	`xai`	`xai` OAuth, device, or API key
Mistral AI	`mistral`	`mistral` API key
Volcano Engine	`volcengine`	`volcengine`, `volcengine-coding`, `volcengine-plan`
BytePlus	`byteplus`	`byteplus`, `byteplus-coding`, `byteplus-plan`
OpenRouter	`openrouter`	`openrouter` dynamic model catalog
Qwen	`qwen`	`qwen`, `qwen-coding-plan`
Z.AI	`zai`	`zai` Global, CN, and Coding Plan profiles
Qianfan	`qianfan`	`qianfan` API key
GitHub Copilot	`copilot`	`github-copilot`, optional `copilot-proxy`
Vercel AI	`ai-gateway`	`vercel-ai-gateway`
OpenCode Zen	`opencode-zen`	`opencode` route with OpenCode Zen auth
Xiaomi	`xiaomi`	`xiaomi` MiMo API
Synthetic	`synthetic`	`synthetic` Anthropic-compatible API
Together AI	`together`	`together` API key
Hugging Face	`huggingface`	`huggingface` Inference Providers route
Venice AI	`venice`	`venice` API key
LiteLLM	`litellm`	LiteLLM proxy URL and key
Cloudflare AI Gateway	`cloudflare-ai-gateway`	built-in Anthropic through Cloudflare AI Gateway path
Custom Provider	`custom`	manual OpenAI-compatible or Anthropic-compatible endpoint

Health And Usage

Agent > Models can show reachability, auth health, discovered models, private-network approval, and known model capabilities. Token and cost accounting stays on Usage, grouped by provider, model, Agent, session, task, and channel.

Overview

Setup and routing

Common providers

Local and private

Additional providers

Model Providers

Model Providers

Model Route

What This Section Covers

Provider Groups

Code-Backed Provider Surface

Health And Usage

​Model Providers

​Model Route

​What This Section Covers

​Provider Groups

​Code-Backed Provider Surface

​Health And Usage

Model Providers

Model Route

What This Section Covers

Provider Groups

Code-Backed Provider Surface

Health And Usage