Interesting model release. It offers ~same capabilities as gpt-oss-120b but with faster inference (should be cheaper as well) and they're also providing (some?) pre-training and (all?) post-training datasets. In some benchmarks it is behind qwen3.5-122b that recently released. Focus on "agentic" stuff, and nvfp4 native. Glad to see nvda pushing for more models, more accessible models, and opening up some of their datasets.
