Get the best builds by email. No account needed.
Live drop
Deep PrincipleDropped Nov 2, 2026
Open-weight 753B MoE post-trained on lab records for chemistry, materials and life-sciences discovery workflows.
About this release
Hydrogen-1.0 is Deep Principle's first open-weight scientific model, created by post-training Z.ai's GLM-5.3 on proprietary domain data from chemistry, materials science and life sciences plus full-cycle lab records and discovery trajectories. It ships under MIT license with a 1M-token context, gated Hugging Face weights, and access via the company's Mira platform or NeuroSpark API. The release targets researchers building experiment-planning, reaction-prediction and structure-elucidation tools.
Read the official Deep Principle launch pageFirst builds incoming — this lane fills in automatically as people start building with Hydrogen-1.0. Know one? Submit the URL.