Hydrogen-1.0: A New Open-Weight Scientific Research Model
Researchers introduced Hydrogen-1.0, an open-weight scientific model built by post-training GLM-5.3 on internal scientific data, lab records, and discovery trajectories. The model demonstrates that domain-specific fine-tuning of powerful open-weight base models can outperform certain proprietary baselines on complex scientific benchmarks. Hydrogen-1.0 achieves a score of 38.3% on the FrontierScience-Research benchmark, outperforming GPT-6-Astra's score of 33.3%.
## BACKGROUND
FrontierScience-Research is a specialized benchmark evaluating AI models on cutting-edge scientific research questions requiring deep domain expertise, multi-step reasoning, and synthesis of complex concepts across disciplines like physics, chemistry, and biology. GLM-5.3 is a large-scale open-weights reasoning model developed by Z.ai, designed for complex reasoning tasks with a 1M-token context window.