LLMTray is a native macOS menu bar app for running open AI models locally on Apple silicon, built on MLX. Your prompts, files and results stay on your Mac.
Chat with open models: a guided setup suggests a model size that fits your memory, and you can download more MLX models from Hugging Face. Vision models read images, reasoning models show their thinking.
Projects with documents: drop in PDF, Word, Markdown or text files and the model searches them and cites the pages it used.
Local images and music: generate and edit pictures right in the chat, or make songs with vocals and lyrics.
For developers: an OpenAI-compatible and Anthropic-compatible API on localhost, so coding agents and other apps can use the same model. Per-model profiles for KV cache quantization and speculative decoding, plus a built-in benchmark.
Every heavy feature is off until you turn it on. No account, no ads, no tracking. Free, open source (Apache 2.0).
No comments or reviews, maybe you want to be first?