[KongchangAI]
ModelHydrogen-1.0

Hydrogen-1.0

An open-weight scientific large model released by Deep Principle for chemistry, materials science, and life sciences. It has 753B parameters, is post-trained on GLM-5.3, and is released under the MIT license.

Timeline (last 90 days)

Nov 2

Deep Principle released Hydrogen-1.0, an open-weight science model for chemistry, materials science and life sciences, with 753B parameters.

Unverified50%
Nov 2

Hydrogen-1.0 was not pretrained from scratch but post-trained on Z.ai's GLM-5.3.

Unverified50%
Nov 2

Hydrogen-1.0's training data includes domain data, laboratory records and agent discovery trajectories.

Unverified50%
Nov 2

Hydrogen-1.0's model weights are released under the MIT License, allowing commercial use and modification.

Unverified50%
Nov 2

Deep Principle claims Hydrogen-1.0 outperforms GPT-6-Astra on the FrontierScience-Research benchmark, but the result lacks third-party reproduction.

Unverified50%
Nov 2

The 753B parameter scale places high demands on deployment hardware, making local inference difficult for most teams.

Unverified50%
Nov 2

The specific composition and cleaning methods of Hydrogen-1.0's training data are not disclosed, which affects outside judgment of the model's capability boundaries.

Unverified50%

All Facts (7)

Source Articles