project-maya

What is project-maya

Project Maya is a locally run tool that supports private deployment of the GLM-5.3-Flash (321B MoE) model on personal GPUs (NVIDIA or AMD, 1 to 16 cards, Linux or Windows). It provides an engine, server, dashboard and APIs compatible with OpenAI and Anthropic, and includes quantisation tools.

Key features

• Supports private deployment of the GLM-5.3-Flash (321B MoE) model on personal GPUs (NVIDIA or AMD, 1 to 16 cards)
• Provides an engine, server, browser dashboard and chat interface, with support for image generation
• Compatible with OpenAI and Anthropic APIs, supports tool calling and concurrent request handling
• Includes a self-developed quantization tool that can quantize models down to 96.5 GB while maintaining high zero-shot accuracy
• Automatically measures GPU, CPU, RAM and SSD and performs self-tuning, with context length support up to 1M tokens

Who it is for

For developers, researchers and local AI application users who need to run large GLM-5.3-Flash models privately on a local GPU. Supports Linux and Windows.

How to get started

1. Visit https://github.com/mw00/project-maya
2. Install following the one-command setup in the README
3. Download quantized model weights such as Maya-S
4. Start the server and browser dashboard to begin using

Compiled from public descriptions on the tool's official website / README / launch page. Pricing and user numbers have not been verified; please check the official website. Some sections were drafted with AI assistance. Written in Chinese and English; other languages are machine-translated.