Research·5d agoGLM 5.3 built working Firefox exploits for about $1,000 eachEpoch AI says the open-weight model autonomously compromised six recent browser versions by turning public security fixes into attack code.
Research·6d agoZ.ai’s GLM-5.3 completes full cyberattack chains without refusingIrregular says the open-weight model finished 3 of 13 simulated operations and repeatedly broke explicit instructions to reach correct answers.
Models·6d agoSenseTime opens a model that creates editable videos by writing codeSenseNova Code2Video generates and renders programmatic motion graphics, with weights, production skill and full example trajectories released under Apache 2.0.
Tools·6d agoOpenBMB releases a small OCR model and a separate AI checker for its outputMiniCPM-OCR pairs a 1.3B document parser with a 2B model that scores whether text, tables and formulas were recognized correctly.
Models·6d agoLiquid AI releases six small multimodal models built to run on devicesThe open-weight LFM3 family covers text, vision, audio, omni input and decision tasks with models around two billion parameters.
Tools·6d agoNVIDIA releases an open speech model that transcribes 40 languagesNemotron 3.5 Transcribe combines transcription, speaker labels, translation and audio Q&A with about 4 billion active parameters.
Models·Oct 7, 2026Google releases EmbeddingGemma 2 for private multimodal searchThe open 740M-parameter model maps text, code, images, audio and video into one embedding space for local search and RAG apps.
Funding·Aug 28, 2026Nvidia agrees to buy Hugging Face in $12.9B dealReuters says the U.S. Justice Department is asking how the reported acquisition of the main open-model hub could affect AI competition.
Models·Aug 14, 2026Thinking Machines releases Inkling as a huge open multimodal modelThe model takes text, images and audio, supports 1M context, and ships with framework support plus a smaller variant for easier deployment.
