The End of LLMs? TypeSafe AI’s Jev Model Ditches Language for Probabilities
Jev, a new transformer model from TypeSafe AI, produces probabilities not language, promising cheap, fast, hallucination-free automation.
Llm
4-Bit LLM Beats Full-Precision: The Compression Tradeoff Just Got Inverted
A new technique called Quantization-Aware Healing (QAH) compresses GPT-OSS 120B to 60B at 4 bits while beating the original on reasoning and math benchmarks — a breakthrough for efficient model deployment.
Alibaba's Open-Source Qwen3.8-Max Dethrones GPT-5.6 on Agentic Benchmarks
Alibaba’s Qwen3.8-Max outperforms GPT-5.6 Sol Max on agentic benchmarks, and the company will release open weights next week—a first for the Max-class line.
Google Ships Three New Gemini Flash Models, Flagship Pro Delayed
Google released three new Gemini Flash models — 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber — offering better efficiency and lower cost. Meanwhile, the much-anticipated Gemini 3.5 Pro continues to be delayed.
China's AI Double Punch: Moonshot and Alibaba Release Frontier Models Rivaling OpenAI
China’s Moonshot and Alibaba released K3 and Qwen3.8, claiming performance near GPT-5.6 and Claude Fable 5 at lower costs, intensifying the US-China AI competition.
Moonshot AI's Kimi K3: 2.8 Trillion Parameters, Open-Weight, and a Challenge to US Hardware Restrictions
Moonshot AI’s Kimi K3 is the largest open-weight model ever released at 2.8 trillion parameters, available under a modified MIT license. It signals that Chinese AI developers are circumventing US hardware constraints and matching frontier-class American models.
Unisound's U2: An Agentic LLM That Actually Finishes the Job
Unisound’s U2 is a new general-purpose LLM built for execution, not just conversation. It autonomously handles complex, multi-step workflows across office work, software engineering, and research.