Its repository describes an engine, server, dashboard and quantization tools, with NVIDIA support and experimental AMD HIP work documented in the project history. The release shifts the deployment option from hosted inference toward operator-managed local execution; it does not establish broad hardware compatibility or production-grade performance across configurations.
Raw SignalOct 12, 20261 min read1 source
Project Maya is an MIT-licensed inference stack that runs the 321-billion-parameter GLM-5.3-Flash mixture-of-experts model on Linux workstations.
In one minute
Project Maya is an MIT-licensed inference stack for running the 321-billion-parameter GLM-5.3-Flash mixture-of-experts model on local Linux hardware.