OpenAI's GPT-5.4 Solves a 20-Year Math Problem, Anthropic Gets Designated a Supply Chain Risk, Qwen Drama Unfolds
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
Weekly AI news roundup covering GPT-5.4 Thinking and Pro release with 1M context, mid-reasoning steering, and a Polish mathematician's claim that it solved a 20-year research math problem. Anthropic refused DoD demands to remove restrictions on autonomous kill chains and citizen surveillance, resulting in being designated a US supply chain risk — the first such designation for a US company — while OpenAI signed a deal with the Department of War. Alibaba's Qwen team released small multimodal models (0.8B–9B), and Qwen tech lead Junyang Lin's cryptic departure tweet went viral, prompting Alibaba CEO intervention. StepFun released Step 3.5 Flash Base, a 196B sparse MoE model under Apache 2.0. Wolfbench.ai was introduced as a four-metric agentic benchmark framework comparing model reliability across harnesses.
