BaseRT is the fastest LLM runtime on Apple Silicon, designed for engineers building on-device AI. It runs open source models locally with one command, offering up to 6.4x faster prefill than llama.cpp and 3.9x faster than MLX, with up to 1.33x faster decode.
Fastest LLM runtime on Apple Silicon
- Running local LLMs for on-device AI development
- Powering coding agents with private, offline inference
- Accelerating model prefill and decode for Apple Silicon users
- Serving open source models for prototyping and testing
- Enabling privacy-preserving AI workflows without cloud dependencies
- Install BaseRT with a single command on your Apple Silicon device
- Serve a supported open source model using BaseRT
- Point your local coding agent or application to the served model endpoint
- Run inference entirely on-device without API keys or data leaving your machine
- Monitor performance metrics like tokens per second
A modern, visual file manager for the terminal
Give Claude Code and Codex a voice
Sync your mailbox with your issue tracker
Privacy-friendly web analytics with traffic and revenue.
Your time travel debugger
The easiest way to deploy a managed agent.
Find & fix live vulnerabilities in Vibe Apps with 1-prompt.
Manage your team of AI agents by voice, from anywhere
Full-Stack Platform for Training Small Language Models
The AI database client for Mac
API integration testing that remembers what breaks
Comment on your AI agent's code & collaborate on changes
A private data-usage meter for your Mac menu bar
Annotate anything for humans and their agents
A terminal, a real browser, and Claude Code under one roof
Screen recordings that edit themselves
The open source Ahrefs alternative
Control every aspect of model training and fine-tuning
