Overview

Canto is a privacy-first AI-powered notebook for macOS that runs entirely on-device. It combines writing, coding, and knowledge management with six local LLM models (Mistral 7B, Llama 3.1, Qwen 3, Gemma 2) accelerated via WebGPU — no cloud, no subscriptions, no data uploads. Features include Memory Links for automatic semantic connections across notes, Jupyter-style code notebooks supporting Python (via Pyodide), JavaScript, and TypeScript, and AI assistance through Cmd+K prompts, agent chat, selection AI, and real-time autocomplete. All notes are stored in an encrypted local SQLite database.

Pricing: $24.99

Minimum macOS: 14.0 (Sonoma)

Architecture: Apple Silicon

Key Features

6 local AI models (Mistral 7B, Llama 3.1 8B, Qwen 3, Gemma 2 9B) with WebGPU acceleration
Memory Links — automatic semantic linking that surfaces related notes as you write
Code notebooks with Python (Pyodide), JavaScript, and TypeScript execution inline
Cmd+K instant AI prompts, Agent Chat for multi-step tasks, Selection AI for quick actions
Real-time AI autocomplete from your notes as you type
Semantic search across all notes with AI embeddings
Encrypted SQLite database — nothing ever leaves your device
Auto-install Python packages (numpy, pandas, matplotlib)
WikiLinks and backlinks for manual note connections
Works 100% offline after initial model download
32K–128K context window support across models
One-time purchase with lifetime updates