On an RTX 5070 Ti 16GB and 32GB RAM, Qwen3.6-35B-A3B Q8_0 achieved 44 t/s.
The White House condemned China's industrial-scale theft of AI technology and signaled a response.
Reports indicate that the Chinese government plans to restrict investment in US tech companies.
Introduced a method to mount tar.gz directly as a filesystem in WebAssembly without extracting it.
Oliveyoung Global Engineering Center held its first workshop, combining a GenAI hackathon with a review of its 3-year roadmap.
Anthropic acknowledged the degradation in Claude Code quality and explained the cause as product-level modifications.
A parser error in vLLM was found to cause malfunctioning Tool Calling functionality in Qwen models.
A man who spread an AI-generated wolf photo was arrested by South Korean police.
Box, a security-enhanced Android AI app that integrates LiteRT, llama.cpp, and more to operate fully offline, has been released.
DeepSeek has released the V4 model family, which supports a 1M token context.
Canadian AI company Cohere has acquired Germany's Aleph Alpha to expand into the European market.
Max-and-Omnis released GGUF files for Nemotron-3-Super-64B-A12B-Math-REAP.
Qwen3.6 35B-A3B showed a prompt processing speed of 282 tokens per second on a Radeon 780M iGPU.
Spinel, an AOT compiler that turns Ruby into native executables, has been released.
MyRealTrip PE Kook Seunghyun voluntarily switched from settlement development to sales.
The top score in the public inference challenge for Tesla T4·Qwen2.5-0.5B is 3,536 tok/s.
GitHub Copilot's documentation now reflects a 0.25x rate for GPT-5.4 nano.
On a combination of m720q i7, 64GB RAM, and T1000 8GB, Qwen 3.6 27B recorded 20 TPS.
DeepSeek unveiled two 1M token context V4 preview models, pushing a low-price strategy.
GPT-5.5 and GPT-5.5 Pro have been added to Vercel AI Gateway.
The US Department of Defense is dramatically increasing its budget for autonomous warfare capabilities and cultivating AI software as a core strategic asset.
OpenAI has released GPT-5.5, a new model with enhanced agentic capabilities and reasoning power.
DeepSeek-V4 was released, featuring 1M-token context and strong coding benchmark results.
Simon Willison introduced a tool that converts milliseconds from LLMs into seconds and minutes.