what matters in AI right now

01Muse Wonder: Meta Beats Opus 5.5 on Long Coding Projects, 57.5 vs 46.3October 12
02SpaceX buys Cursor for $60 billion; Gergely Orosz doubts top engineers will followOctober 12
03agent-skills: 25 skills make the agent prove its code works before it says 'done'October 11

Loading the page.

October 10, 2026

New Models Fail to Replicate Human Algorithmic Invention: Epoch AI InnovationEval Test

Epoch AI published InnovationEval on Oct 9, 2026: new models failed a task that required inventing an algorithm rather than solving a known one.

Source

One email a week - only what changes how you work. Subscribe

what matters in AI right now

01Muse Wonder: Meta Beats Opus 5.5 on Long Coding Projects, 57.5 vs 46.3October 12
02SpaceX buys Cursor for $60 billion; Gergely Orosz doubts top engineers will followOctober 12
03agent-skills: 25 skills make the agent prove its code works before it says 'done'October 11