https://openmodelweights.com/news/microsoft-mage-vl-cowork-4b-hybrid-computer-use-20261101/Open Model Weightsen2026-11-01T13:44:07ZMicrosoft opens Mage-VL-Cowork: a 4B agent that switches between screen and shell https://openmodelweights.com/news/nvidia-deepseek-v41-flash-1017-nvfp4-blackwell-20261101/Open Model Weightsen2026-11-01T13:44:07ZNVIDIA packs DeepSeek V4.1 Flash 1017 into a 400 GiB Blackwell-oriented NVFP4 release https://openmodelweights.com/news/surya-25-open-ocr-689m-faster-document-parsing-20261101/Open Model Weightsen2026-11-01T13:44:07ZSurya 2.5 releases 689M OCR weights with faster document parsing and stricter licensing https://openmodelweights.com/news/nvidia-metatron-v1-open-humanoid-policy-20261031/Open Model Weightsen2026-10-31T10:54:30ZNVIDIA opens Metatron v1: one G1 humanoid policy for movement, interaction and recovery https://openmodelweights.com/news/transformers-5220-mistral-large4-fp8-adapters-20261031/Open Model Weightsen2026-10-31T10:54:30ZTransformers 5.22 brings Mistral Large 4 support and makes FP8 adapter training more practical https://openmodelweights.com/news/ollama-04013-lfm3-audio-omni-linux-windows-20261031/Open Model Weightsen2026-10-31T10:54:30ZOllama 0.40.13 takes LFM3 audio and omni inference beyond Apple Silicon