A Linux Foundation study finds generative AI brings large productivity gains to open-source development. However, about a quarter of the change proposals it produces have little or no value.
Tsinghua researchers have released AKV, a serving system built on SGLang that lets LLM agents drop and reposition parts of their KV cache rather than summarise their history. Across four agent benchmarks, it cut average cost by 27.7%.
Nanbeige LLM Lab has released Nanbeige4.3-33B, a 33-billion-parameter dense model under Apache 2.0 with a 512K context window. It scores 80.9 on Terminal-Bench 2.1 and will be used to teach the forthcoming Nanbeige4.3-3B.
At PyTorch Conference North America 2026, Google and Meta released TorchTPU, a native PyTorch backend for Google's TPUs, and Meta said it plans to shift larger LLM workloads onto TPUs with it. AWS released TorchNeuron for Trainium the same week, while developers questioned TPU availability and where vendor lock-in now lies.
Google has open-sourced ML Drift, a cross-platform GPU compute engine for on-device AI and machine learning inference, under Apache 2.0. It hides the differences between GPU hardware and low-level APIs and supports OpenGL ES, OpenCL, Metal and WebGPU.
Hugging Face's Carbon-X is an open 14.8-billion-parameter DNA language model built on Qwen3-14B. It reads raw DNA in a chat and answers in English, averages 66.7% across five DNA benchmarks, and is released under Apache 2.0.