Skip to content

Provider Catalog

The desktop app ships a catalog of 34 providers (source of truth: NativeProviderId in process/providers/types.ts plus the catalog UI; one entry is currently hidden pending full wiring). You connect each one in Settings -> Models by pasting an API key or, for the cloud providers, filling the credential form. Five providers are also native to the Wayland Core engine, which means the CLI can talk to them directly without the desktop. Everything else in the catalog routes through the engine using an OpenAI-compatible layer (ProviderCompat).

Flux Router is featured at the top of the Settings -> Models screen. Every connected provider has a per-provider on/off toggle so you can disable it without deleting its credentials.

The provider id is the value the app stores internally and is the same id you use when you connect through the API. The group column matches the Browse grouping in the Models panel.

These five work in both the desktop and the standalone wayland-core CLI. They have first-class auth paths in the engine.

ProviderIdAuth
AnthropicanthropicAPI key (Claude subscription via the Claude Code CLI)
OpenAI (and OpenAI-compatible)openaiAPI key, optional base URL
Google Geminigoogle-geminiAPI key (native Generative Language API)
AWS Bedrockaws-bedrockAWS SigV4 (access key, profile, or environment)
Google Vertex AIvertexGCP OAuth2 or service account

Local models reach the engine through the OpenAI-compatible path. See Use a Local Model.

All 34 providers, in the catalog’s render order. The five engine-native providers are marked.

ProviderIdGroupEngine-native
AnthropicanthropicFrontierYes
OpenAIopenaiFrontierYes
Google Geminigoogle-geminiFrontierYes
AWS Bedrockaws-bedrockCloudYes
Google Vertex AIvertexCloudYes
OpenRouteropenrouterOpenNo
GroqgroqOpenNo
xAI GrokxaiFrontierNo (supports keyless OAuth, see Sign in with X)
MistralmistralFrontierNo
CoherecohereFrontierNo
PerplexityperplexityOpenNo
Together AItogetherOpenNo
Fireworks AIfireworksOpenNo
CerebrascerebrasOpenNo
ReplicatereplicateOpenNo
Hugging FacehuggingfaceOpenNo
NVIDIA NIMnvidiaOpenNo
AnyscaleanyscaleOpenNo
DeepSeekdeepseekChineseNo
MoonshotmoonshotChineseNo
QwenqwenChineseNo
BaichuanbaichuanChineseNo
YilingyiwanwuChineseNo
Zhipu GLMzhipu-glmChineseNo
MiniMaxminimaxChineseNo
Sakana AI (Fugu)sakanaFrontierNo
Stability AIstabilityOpenNo
DeepgramdeepgramVoiceNo
AssemblyAIassemblyaiVoiceNo
ElevenLabselevenlabsVoiceNo
Azure OpenAIazureCloudNo
Flux Routerflux-routerOpenNo
OpenAI-compatibleopenai-compatibleOpenNo
Ollama (Local)ollama-localOpenNo

The chatgpt-subscription provider connects your ChatGPT Plus, Pro, Team, or Enterprise subscription via a one-click OAuth flow. It does not appear in the main provider grid because it has no API key path. Instead, use the Sign in with ChatGPT button on the ChatGPT card in Settings -> Models to activate it. Once connected, a static set of GPT-5 and Codex models (gpt-5.2-codex, gpt-5.2, gpt-5.1-codex-max, gpt-5.1-codex, gpt-5.1-codex-mini, gpt-5.1, codex-mini-latest) becomes available in the model picker. See Sign in with ChatGPT for the full walkthrough.

  • Frontier and Open providers connect with a single API key. The app recognizes most keys by their prefix and selects the right provider for you.
  • Cloud providers (AWS Bedrock, Google Vertex AI, Azure OpenAI) use a multi-field credential form rather than a single key.
  • Voice providers (Flux Voice, Deepgram, AssemblyAI, ElevenLabs) power speech-to-text and text-to-speech. See Set Up Voice.
  • Stability AI and Flux Router are the providers most often paired with image generation. See Set Up Image Generation.
  • Sakana AI (Fugu) is a first-class provider you connect with an API key. Its Fugu models route through the OpenAI-compatible path, so they appear in the model picker once the provider is connected.
  • OpenAI-compatible is the entry to use for any server that speaks the OpenAI API, including a local Ollama or vLLM endpoint.