Deep Principle’s open-weight Hydrogen-1.0 learned from lab notebooks and says it beats GPT-6-Astra at research science. The scores are its own.
Tag: training data
Google, Microsoft, OpenAI and Anthropic will read one robots.txt line asking them not to train on a site. It only looks forward.
Nvidia’s open-weight Nemotron 3.5 Transcribe handles transcription, translation and speaker labels from one prompt. The scores are self-reported.
Moonshot’s Kimi K3.1 Pro nears the top closed models on coding, by its own numbers. Before the open weights ship, it removed the hacking lessons.
Ai2’s open FairStress benchmark tests whether AI models hold fair decisions under pressure. The ones tested bent more easily one way.
Microsoft finished pre-training MAI-Base-2, a 2.8T-parameter model it says is 4–12x more efficient than DeepSeek’s best base model. Not usable yet.
Swiss AI’s Apertus 1.5 adds images, audio and a 262k context. Now every training stage, dataset and script behind it is public too.
Ai2 released AstaBrief 8B, an Apache-licensed model that turns a research question and paper excerpts into a cited report in one pass.
A new arXiv paper labels 31.1% of filtered August web tokens as AI-written — then shows, across 800 models, when those tokens stop helping.