11 providers, 70+ models
Anthropic, OpenAI (incl. the GPT-5.6 Responses API), Google Gemini, xAI Grok, DeepSeek, Mistral, Kimi, Qwen, Perplexity, Cerebras, DeepInfra. Live cost tracking per message.
The local-first AI workspace. Every frontier model — Claude, GPT, Gemini, Grok, DeepSeek, Mistral, Kimi, Qwen — plus local MLX models, in one native macOS window.
One window, eleven providers, one keyboard shortcut away.
Built native in SwiftUI — no Electron, no lag, no telemetry.
Anthropic, OpenAI (incl. the GPT-5.6 Responses API), Google Gemini, xAI Grok, DeepSeek, Mistral, Kimi, Qwen, Perplexity, Cerebras, DeepInfra. Live cost tracking per message.
A curated library of 20 mlx-community models with one-click Hugging Face downloads. The server starts on demand — private, offline, free.
Web search, file access and MCP servers mid-conversation, streaming around results — with Auto / Ask / Off permission per tool.
Attach screenshots and photos — sent as native multimodal blocks to every vision-capable provider.
Think→tool→observe loops with 18 built-in tools, folder scoping, per-call approvals — streamed live as they work.
Edit & resend any prompt, fork conversations at any message, regenerate any answer, clickable source citations.
On-device AES-GCM memory with semantic retrieval — profile facts, project context, folder-scoped knowledge.
Substantial outputs become versioned artifacts with manual and AI editing, HTML/Markdown preview and export.
A global hotkey summons a floating chat over any app — ask, copy, or paste the answer right where you were typing.
Open the DMG and drag API Chat into your Applications folder.
Paste any provider key, or skip keys entirely and download a local MLX model.
Free download. Your keys, your data, your Mac.
⬇ Download API Chat 1.7.0