A new agents paper takes aim at a specific failure mode: recursive self-improvement loops that rewrite an LLM agent's harness end up memorizing ...
Tencent's ARC Lab published GameHorizon Suite on Sept 21, comprising an automated multi-horizon instruction annotator, a 5,000-hour AAA gameplay ...
Tencent released a new image model on the same day Alibaba's Apsara conference opened. Its shares gained more than 7% in Hong Kong. The product, Hy ...
A Sept 21 arXiv paper from Peking University introduces Harness-Zero, an 'agent-as-harness' training loop that distills behaviors from ...
OpenAI says an internal model has resolved more than 100 additional open problems across most areas of mathematics, and it has recruited ...
Apple asked the court for direct access to forensic images from ex-employee Chang Liu's devices, saying its experts identified 'critical flaws' in ...
Alibaba CEO Eddie Wu used his Apsara keynote in Hangzhou to plan a 5–10T-parameter successor to Qwen (currently ~2.4T) and unveil the Zhenwu V900 AI ...
Amazon says its engineers watched Perplexity's servers hit Amazon's servers every day between 18 March and at least 11 May of this year, according to ...
Nvidia has certified its first outside partners for the physical guts of an AI factory. In a company blog post on Sept. 21, the company introduced ...
Zelda Williams posted on X on 22 September calling out a viral 'private video' of her late father Robin Williams as unconvincing AI, adding 'leave him ...
IntBMoE, a Mixture-of-Experts architecture described in an arXiv preprint posted September 18, is already running in AMap's generative ...
A frozen frontier model, given a growing notebook of natural-language skills, drove Claude-Sonnet-4's GenEval2 execution success from 72.7% to ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results