Skip to content

Providers

Foysal Ahammad edited this page Oct 3, 2026 · 3 revisions

🌍 Providers & Models

Connect with /connect, switch models with /model.

🏢 Supported out of the box

OpenCode Zen · Pollinations (free) · OpenAI · Anthropic · Google · Groq · DeepSeek · Mistral · xAI · Together · OpenRouter · HuggingFace · GitHub Models

✨ Per-model capabilities

Capability Meaning
🔧 Tools can call the agent's tools
🖼️ Images accepts image input
🧠 Reasoning streams thinking/reasoning
⚡ Streaming token-by-token replies

/model shows these flags for every model of the connected provider.

🛟 Resilience

  • automatic retry with exponential backoff (429 · 5xx · timeouts)
  • provider fallback when an endpoint is down
  • a down free-tier model is answered via a healthy fallback model — the reply says so in a notice line

🔑 API keys stay on your device, never appear in chat, and are never written into exported sessions.


Clone this wiki locally