Shane Melaugh argues that Claude Opus 5.5’s reasoning behavior changed abruptly on November 3, even though its model name remained the same. He says he had Gemini 4 Argon examine 1,400 Claude Code sessions and found a sharp shift in how the model reasoned. In a follow-up, he adds that Opus 5.5 in “high” mode began using substantially less reasoning. The claim is that this change could explain why some users feel the model has become less capable, rather than their experience being imagined.
Melaugh also describes a test on a real software bug: Opus 5.5 in high mode reportedly started acting too quickly, became confused, and failed to find a fix after three hours. In contrast, the “xhigh” setting paused to think and solved the bug within minutes. He concludes that the advice to use high mode for everything may no longer hold. Theo, who had popularized that approach, asks for an independent investigation using Claude Code logs stored in ClickHouse, signaling that the reported change still needs verification beyond Melaugh’s analysis and example.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Democrats’ gains in the 2026 elections could bring new scrutiny of AI, datacenter expansion, and Elon Musk when the new Congress takes office in 2027. Political backlash against these issues is growing. (theregister.com)
samir-vllm is a small LLM inference engine built from scratch. The repository includes over six hours of video tutorials and 85 pages of handwritten notes. (github.com)
Opting out of Google’s AI search removed the site from AI Overviews and AI Mode within an hour, wiping out 500-1,000 daily impressions - about 10% of its search impressions. The site kept its top organic ranking, but links in AI answers had a 0.9% click-through rate; visibility returned quickly after opting back in. (signal.zyppy.com)
South African author Coral Hart has published more than 370 books in 20 months using several AI models and a detailed process she designed. Despite online abuse and death threats, she says AI is here to stay and that its writing has improved rapidly. (enca.com)
Next-Concept Modeling generates text by predicting hierarchical discrete codes in parallel, moving from coarse representations to finer details. It runs 10-31 times faster than strong diffusion baselines on text benchmarks and outperforms autoregressive and diffusion models on Sudoku reasoning. (arxiv.org)
Unreviewed AI-generated work shifts the burden and risk to teammates. Developers who simply relay an AI agent’s output should face discipline, the author argues. (swiftrocks.com)
Today's best Hacker News stories, summarized and screenshotted, one email a day.