Gensyn releases open-1b, the first fully auditable and replayable foundation model

flngr · x · 2026-09-17

Gensyn has released its first foundation model, open-1b. Unlike models from other AI labs that require trusting their training process, open-1b is fully auditable and replayable: you can take any checkpoint, re-run it with their custom kernels, and get a bit-identical result across NVIDIA and Apple hardware.

It builds on compiler, kernel, and training technology the team has developed for years. Their argument: if the world is going to offload more of its intelligence to machines, those machines must be verifiable — open-1b is a milestone toward verifiable AI.

Related event: Gensyn releases open-1b, the first fully auditable open-source model(8 posts)→

Original post →

More from Models

Models channel →