project-maya: run 700B-class open models on your own GPUs
project-maya is an open-source runner for the newest giant open-weight models — GLM-5.3 (744B), GLM-5.3-Flash (321B), DeepSeek-V4.1-Flash (748B) and MiniMax-M3.1-Flash (428B) — on the user's own GPUs, and it is gaining about 300 stars a week.
Why it matters
The newest 700B+ open models are unusable without a serving stack: project-maya packages them into a local runner so a multi-GPU owner can actually use the current frontier at home.
What you could build with it
A researcher with a multi-GPU workstation could run a 744B open model locally through project-maya for private experiments on sensitive data, keeping everything on their own hardware instead of sending it to a cloud API.
Does it hold up?
Usable today for owners of multi-GPU rigs: runs 300-700B+ open models locally with documented hardware needs; 2,794 stars and roughly 300 gained in a week show real adoption, but setup is DIY and hardware-gated.
Built with project-maya
- Best Local & Private AI Tools on GitHub (November 2026)article · Daily ranking by measured star velocity lists project-maya among the fastest risers for running huge open models locally.
- Project Maya: run large MoE LLMs on your own GPUsarticle · Directory deep dive: the four supported MoE families, the GPU-RAM-NVMe expert tiering, and the v1.1.65 stability release of 16 November with AMD/Windows start-up fixes.
- Project Maya Lets You Run 700B-Parameter MoE Models on Your Own GPUsarticle · Coverage of the project's pitch — GLM-5.3-Flash and DeepSeek-V4.1-Flash-1017 privately on owned GPUs — with OpenAI/Anthropic-compatible APIs and scaling from one card to sixteen across NVIDIA and AMD.
Learn more
First spotted on github: source.
More AI models
VoiceStudio: open-source, fully-local ElevenLabs alternative for voice cloning, dubbing, and audiobooksdebpalash/VoiceStudio is a desktop app that runs voice cloning, voice design, video dubbing, dictation, transcription…models · JEV 0.75Edge0-145B: open 145B MoE that runs on a single consumer GPUEdge0 open-sourced 145B and 120B latent-MoE models derived from Qwen3.8-Flash-Next, plus an inference framework that…models · JEV 0.74Strata: run a 125B model on a gaming PCFree, MIT-licensed local inference engine that runs Qwen3.8-Flash-Next (125B MoE, about 6B active per token) on…models · JEV 0.73Google AI Edge Gallery: offline speech recognition and synthesis on-deviceGoogle added fully offline speech recognition and text-to-speech to AI Edge Gallery, with ML Kit APIs on Android and…models · JEV 0.73Soup: fine-tune an 8B LLM on a 4 GB laptop GPU from one YAML fileAn open-source CLI that fine-tunes LLMs from a single YAML file, with layer streaming that trains an 8B model on a 4…models · JEV 0.73Tencent Hunyuan Hy3: open 295B MoE with 21B active under Apache 2.0Tencent open-sourced Hy3, a 295B-parameter mixture-of-experts model with only 21B parameters active per token, under…models · JEV 0.73
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.