One extension, every AI provider
Pick the AI that fits your task and budget. Switch providers mid-session, mix free and paid models, or run everything locally with Ollama. Your keys stay in your OS keychain.
At a Glance
| Provider | Auth | Transport | Free? |
|---|---|---|---|
Claude (Anthropic) | API key | Agent SDK | Paid |
MR Coder (Router) | OAuth connect | Agent SDK | With subscription |
Kilo Code | CLI OAuth | Subprocess | Free models |
Qwen (DashScope) | API key or CLI | SSE / Subprocess | Free trial |
OpenRouter | API key | HTTPS SSE | Free models |
Ollama (Local) | None | Localhost | Free forever |
OpenAI Codex | API key or CLI | SSE / Subprocess | Paid (free trial) |
Google Gemini | API key or CLI | SSE / Subprocess | Free tier |
HuggingFace | HF token | HTTPS SSE | Free tier |
Provider Details
Claude (Anthropic)
PaidMost capable reasoning model for complex coding tasks
Top Models
- Claude Opus 4.8200K
- Claude Sonnet 4.6200K
- Claude Haiku 5.51M
Setup
Get API key from console.anthropic.com → Set Claude API Key command
MR Coder (Router)
With subscriptionClaude models via managed router — connect your mr-coder.io account and skip API key setup
Top Models
- Claude Opus 4.8200K
- Claude Sonnet 4.6200K
- Claude Haiku 5.51M
Setup
Connect Account command → approve on mr-coder.io → provider auto-configures
Kilo Code
Free modelsFree models including Grok and Nemotron via CLI
Top Models
- Grok Code Fast256K
- Nemotron 3 Super 120B128K
- Step 3.5 Flash262K
Setup
npm install -g @kilocode/cli → kilo signin → Pick Kilo Code provider
Qwen (DashScope)
Free trialAlibaba's coding models — API or CLI, with free trial
Top Models
- Qwen3 Coder Plus131K
- Qwen3 Coder Flash131K
- Qwen Turbo128K
Setup
Get DashScope API key → Set Qwen API Key, or: npm install -g @qwen-code/qwen-code
OpenRouter
Free modelsFree models from multiple providers, one API key
Top Models
- Nemotron 3 Super 120B128K
- GPT-OSS 120B128K
- MiniMax M2.5196K
Setup
Sign up at openrouter.ai → Generate key → Set OpenRouter API Key
Ollama (Local)
Free foreverRun models 100% locally — free forever, no API key needed
Top Models
- Qwen 3.6 35B256K
- Qwen 2.5 Coder 7B32K
- Llama 3.3 70B131K
Setup
Install Ollama → ollama pull qwen3.6:35b → Pick Ollama provider
OpenAI Codex
Paid (free trial)GPT-4.1 and o4-mini with 200K+ context for complex refactors
Top Models
- o4-mini200K
- GPT-4.11M
- Codex Mini200K
Setup
Get key from platform.openai.com → Set OpenAI Codex API Key
Google Gemini
Free tier1M context window, generous free tier — no credit card required
Top Models
- Gemini 2.5 Pro1M
- Gemini 2.5 Flash1M
- Gemini 2.0 Flash1M
Setup
Free key from aistudio.google.com (no credit card) → Set Gemini API Key
HuggingFace
Free tierOpen-weight coding models — free tier with HF token
Top Models
- Qwen3 235B128K
- DeepSeek V3128K
- Llama 4 Scout 17B131K
Setup
Free token at huggingface.co/settings/tokens → Set HuggingFace API Token
Mix & match, mid-session
Click the provider chip in the chat header to switch providers or models at any time. Use Claude for complex architecture, Ollama for quick local edits, and Gemini for large-context refactors — all in the same workspace.
Always Free
Ollama runs locally with no API key. Kilo Code offers free models via CLI. HuggingFace has a free tier. No credit card needed for any of them.
Your Keys, Your Control
All API keys are stored in your OS keychain via VS Code SecretStorage. Keys never leave your machine, never appear in logs or settings files.
Provider Lock per Session
Each chat session is locked to one provider. Switching providers starts a fresh session with an optional "carry summary" to preserve context.
Ready to route smarter?
Join developers who use MR Multi Model AI Coder to access every AI model through one API and one VS Code extension. Free to get started.