Providers
Supported providers and how to connect them.
Arctic supports multiple provider types so you can keep all models in one CLI.
How to connect
- OAuth/device login:
arctic auth login - API keys: set environment variables and launch Arctic
For credential management, see Authentication.
Provider types
Coding‑plan providers
These are subscription‑based coding plans (handled via OAuth/device login):
- Anthropic
- Codex (ChatGPT)
- Google Gemini
- GitHub Copilot
- Z.ai
- Antigravity
API providers
These are API‑key or custom endpoint providers:
- Anthropic
- OpenAI
- OpenRouter
- OpenAI‑compatible endpoints (custom URLs)
Custom providers (OpenAI‑compatible)
You can add OpenAI‑compatible endpoints in config:
{
"provider": {
"my-provider": {
"api": "https://api.example.com/v1",
"npm": "@ai-sdk/openai-compatible",
"options": {
"apiKey": "$MY_PROVIDER_KEY"
}
}
}
}Full list
Show all supported providers
- aihubmix
- alibaba
- alibaba-cn
- amazon-bedrock
- azure
- azure-cognitive-services
- bailing
- baseten
- cerebras
- chutes
- cloudflare-ai-gateway
- cloudflare-workers-ai
- cohere
- cortecs
- deepinfra
- deepseek
- fastrouter
- fireworks-ai
- github-models
- google-vertex
- google-vertex-anthropic
- groq
- helicone
- huggingface
- iflowcn
- inception
- inference
- io-net
- kimi-for-coding
- llama
- lmstudio
- lucidquery
- minimax
- minimax-cn
- mistral
- modelscope
- moonshotai
- moonshotai-cn
- morph
- nebius
- nvidia
- ollama
- ollama-cloud
- opencode
- ovhcloud
- perplexity
- poe
- requesty
- sap-ai-core
- scaleway
- siliconflow
- siliconflow-cn
- submodel
- synthetic
- togetherai
- upstage
- v0
- venice
- vercel
- vultr
- wandb
- xai
- xiaomi
- zai-coding-plan
- zenmux
- zhipuai
- zhipuai-coding-plan