Newswire
Hugging Face
July 2026
- Leanstral 1.5: Mistral's AI model for formal proofs is open source
- Z.ai's open-weights GLM-5.2 beats GPT-5.5 on multiple long-horizon coding benchmarks for 1/6th the cost
- NVIDIA and Hugging Face Bring New Models and Frameworks to LeRobot
- NVIDIA brings Isaac GR00T model and Teleop framework to LeRobot
- Tencent open-sources Hy3 295B MoE AI model under Apache 2.0
June 2026
- DeepSeek Releases DSpark: Speculative Decoding Makes V4 Up to 85 Percent Faster
- BitRobot and Hugging Face Drop HIW-500: A Massive 10TB Real-Home Humanoid Dataset
- Liquid AI releases LFM2.5-Embedding-350M and LFM2.5-ColBERT-350M retrieval models
- Weibo researchers claim 3B-parameter model matches larger AI systems on demanding math benchmarks
- Moonshot AI ships Kimi K2.7-Code targeting token efficiency in agentic coding
- Google DeepMind Introduces Gemma 4 12B Open-Weight Multimodal Model
- Google's Gemma 4 12B unified multimodal model released, runs on 16GB laptop with native audio
April 2026
- SenseTime releases open-source image model that runs on Chinese chips
- Nvidia launches Nemotron 3 Nano Omni multimodal model for agentic AI
- DeepSeek releases V4 Pro and V4 Flash preview models at fraction of frontier prices
- OpenAI releases open-source Privacy Filter for on-device PII redaction
- Google's Gemma 4 E2B Model Adds Audio Processing Capabilities via MLX
March 2026
- Alibaba's Qwen3.5-Omni Is Proprietary, Marking a Shift Away From Its Open-Source Strategy
- Alibaba Releases Qwen3.5-Omni, Claims Omnimodal LLM Surpasses Gemini 3.1 Pro on Audio Benchmarks
- Rakuten AI 3.0 Now Available, Japan's Largest High-Performance AI Model Developed as Part of the GENIAC Project
- NVIDIA Expands Open Model Families to Power the Next Wave of Agentic, Physical and Healthcare AI
- New NVIDIA Nemotron 3 Super Delivers 5x Higher Throughput for Agentic AI
- vLLM Adds Support for NVIDIA Nemotron 3 Super for Multi-Agent AI