What is Recursechat?

RecurseChat is a macOS application that runs large language models locally, enabling offline, subscription‑free interaction with AI. The app features a floating chat window that can be summoned from any application with ⌘+⇧+Enter, remaining always on top for quick access.

Users can chat with PDFs, markdown, and text files using on‑device retrieval‑augmented generation (RAG) powered by embedding and vector search, ensuring file content never leaves the machine. RecurseChat supports a range of model formats, including GGUF, Hugging Face, and OpenAI‑compatible endpoints, as well as cloud services such as Claude and ChatGPT, allowing seamless switching between local and remote models.

It also permits importing ChatGPT conversation history to maintain continuity across local and cloud sessions. The interface keeps user data confined to the device, catering to developers, researchers, writers, and privacy‑conscious individuals who require quick, reliable AI assistance directly on their Mac.

Recursechat pricing Paid

One-time purchase $19.99

One-time purchase $29.99

Verify on the official pricing page.

View plans

Recursechat user reviews

Would you recommend Recursechat?

Recommend this tool?

Recursechat's key features

Offline local AI models
No subscription, one-time purchase
Floating chat global shortcut
On-device RAG with vector search
Supports GGUF, Hugging Face, OpenAI
Import ChatGPT conversation history

Recursechat use cases

Create an offline personal research assistant that scans and queries your PDF notes with RecurseChat’s on‑device vector search, enabling instant answers without internet access
Develop a multi‑model chatbot for developers that can switch between a lightweight local model and a cloud‑based LLM on demand, all within a floating chat window on macOS
Build a private knowledge base for a law firm by importing case files into RecurseChat’s file‑based RAG, allowing lawyers to retrieve case precedents quickly while keeping data fully local