Loose posts

Claude identified a hidden repeat pattern in raw phage DNA and found a previously unknown enzyme system with CRISPR-like repeats nearby, per a Chinese post.

A Zhihu review finds Grok 4.7's median logic score rose to 60.97 and hallucination control is top-tier, but tasks often use 90K-150K tokens and it leans on brute force.

A programmer uses the free AI research workbench ScienceBuddy to search PubMed for the original papers behind a viral Xiaohongshu coffee claim, showing its trace panel.

Claude read literature and genome data to form a hypothesis, and researchers validated a previously unknown enzyme system in the lab.

Opus 5.5 scores 66 on the Artificial Analysis Coding Agent Index, a new high with gains on all three evaluations, but its cost per task rises to $13.04.

The Information reports that two Chinese AI companies are accused of forwarding user requests, including police surveillance footage, to Anthropic's Claude and returning the answers.

Opus 5.5 completed a 29-second animation in Motion in a single pass, telling the story of a worker bee that produces about 1/12 of a teaspoon of honey in her lifetime.

Higgsfield showed Claude Opus 5.5 reconstructing an athlete's long jump as a 4D visualization that makes time a visible dimension.

Claude Opus 5.5 and Higgsfield stacked the frames of one video into a time cube, so an entire gallop is visible at once.

A single index.html built with Canvas 2D and no images, video or libraries renders an endless space journey seen through a spaceship window.

Anthropic says CEPI, WHO's Africa office and Kinshasa's national biomedical institute are using Claude to speed up their response to an unusual Ebola variant.

The benchmark uses 22 hours of newly recorded conversations with over 280 speakers and human-verified labels from head-worn mics; Nvidia's Nemotron 3 leads at 14.72% DER.