NVIDIA has released two new quantized text-generation models: GLM-5.2-NVFP4 and MiniMax-M3-NVFP4. The GLM-5.2-NVFP4 model, based on ZAI's GLM-5.2, is MIT-licensed and available for both commercial and non-commercial global use. The MiniMax-M3-NVFP4 model supports text, image, and video inputs, boasts a 1 million token context window, and utilizes NVFP4 via Model Optimizer, though it is restricted to non-commercial use. AI
IMPACT NVIDIA expands its quantized model offerings, potentially improving accessibility for certain AI applications.
RANK_REASON Frontier-lab model release with system card [lever_c_demoted from frontier_release: ic=2 ai=1.0]
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →