AI Models
High-intelligence, ultra-compact coding models built to run locally. Domain-expert AI that gets smarter, not bigger.
Coaade Code 2.0
Our first AI coding model - highly compressed, built specifically for code generation and understanding. Small enough for any laptop, sharp enough for real work.
10x Compression
Advanced compression reduces model size by up to 10x while maintaining 95%+ quality.
CPU & GPU Support
Runs on CPU out of the box. Leverage your GPU for even faster inference.
Code-First Training
Trained specifically for code generation, completion, refactoring, and explanation across 50+ languages.
Sub-100ms Inference
Optimized for local hardware. Get code completions in milliseconds, not seconds.
Capabilities
Coaade Mind 1.0
The decisive personal intelligence that lives on your device.
A new product line - a Personal Intelligence, not an engineering model. A deliberately small 1.5B model with a 1,000,000-token context that runs fully on-device, from phone to desktop, powered by StrataFlow. Best-in-class at tool calling, decisions, and becoming your AI company brain across Notion, Gmail, and the rest of your data.
Expert tool calling
Native function calling plus full MCP-client support, parallel and agentic loops. It does not just suggest - it acts.
Decisive decisions
Frames options, scores criteria, and commits to a recommendation with the tradeoffs on the table. No endless 'it depends'.
Your AI company brain
Pulls Notion, Gmail, Slack, Drive, and your CRM into a million-token context for grounded, cross-source synthesis.
100% on-device
Runs on Apple, Windows, and Ubuntu - phone to desktop. Private by design, works offline, free after download.
Coming Next
Coaade Code 3.0
coming soonNext generation coding model. Significantly smarter, still ultra-compact.
Coaade Reason 1.0
coming soonReasoning model for complex problem solving and code architecture.
How to Install
Models are installed and managed through the Coaade Desktop App. Download the app, install models with one click.
Download →