Is Cotabby free?
Yes — genuinely. Cotabby is free and open source (AGPL-3.0), runs on-device with llama.cpp or Apple’s Foundation Models, needs no account, and lets you import custom GGUF models. Shadowtype is also free and open source — it’s for people who want a more finished, notarized release.
They’re both free — why pick Shadowtype?
For polish and stability. Cotabby is a beta (v0.4.x). Shadowtype is a stable, notarized release with a homegrown auto-updater, a curated tuned-model catalog, built-in personalization and selection rewrite, and no daily cap — also free and open source.
Will I keep the on-device, offline experience?
Yes. Like Cotabby, Shadowtype runs a local LLM with llama.cpp and Metal on Apple Silicon, works fully offline, needs no account, and sends zero telemetry. Tab-to-accept and system-wide completion behave the same — you’re trading beta tinkering for a finished build, not giving up local-first.
Does Shadowtype need Screen Recording like Cotabby?
No. Cotabby requires Accessibility, Input Monitoring and Screen Recording. Shadowtype’s screen-aware context is optional — completion works on Accessibility and Input Monitoring alone, and you only grant Screen Recording if you want OCR context. One fewer mandatory permission.
Can my own scripts, editors, and agents use the same local model?
Yes. Shadowtype exposes an OpenAI-compatible HTTP endpoint on 127.0.0.1 with no key required, so your terminal, editor plugins, and AI agents can call /v1/chat/completions against the same on-device model. A built-in MCP server plugs into Claude Code and other MCP clients, and the BYOM picker loads any GGUF you drop in. Everything is bound to localhost — nothing leaves your Mac. Cotabby doesn’t offer a local API or MCP server.