mw00/project-maya v1.1.39 is now available as a GitHub release. The release summary dated October 29, 2026 lists system prompt reuse, improved SSD expert reads and model import as the changes. This is a new version of the project that runs large models on a personal GPU. The release source is https://github.com/mw00/project-maya/releases/tag/v1.1.39.
According to the repository description, Project Maya is a tool that runs GLM-5.3-Flash (321B MoE), DeepSeek-V4.1-Flash-1017 (748B MoE) and MiniMax-M3.1-Flash (428B MoE) privately on its own GPUs. It supports anything from a single GPU to 16, targeting NVIDIA or AMD GPUs and Linux or Windows. In addition to the engine, server and dashboard, it also provides OpenAI- and Anthropic-compatible APIs and quantization tools. In other words, a single project covers not only running models but also service connectivity and quantization tooling.
However, the provided information names only the three changes and gives no specific scope or performance figures for the system prompt reuse and the SSD read improvements. No license information is given either. It is therefore not possible to judge from this information alone how much this version changes real processing performance, or which configuration each change applies to.