Unverified50% confidenceEventExact time
A team of 1000 Fable 5.2 agents completed a full Lean formalization of Fermat's Last Theorem in 30 hours; under the same configuration, Opus 5.5 completed only 26% in 18 hours and made almost no further progress by 30 hours.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/9/2026
First Seen
Valid until: 2/7/2027
Sources
Related Entities
Related Claims
UnverifiedIn preliminary multi-agent experiments given a full 8 hours, the programming benchmark peaked at only 30.04 (versus 20.39 for single-agent), and the science-category agent benchmark reached only 31.2, lagging behind a competitor's 51.8.69% similarUnverifiedA civil engineer posted on Reddit describing that after deep use of agentic AI (such as Claude Opus 5.5 and Fable 5.1), work efficiency improved substantially, with tasks that originally took weeks now completed in hours.68% similarUnverifiedIn the synthetic experiment (each scheme run 5 times), the 95th percentile of task completion time dropped from 29.5 minutes to 15.7 minutes.67% similarUnverifiedRAMP had Claude Fable 5.1 run a pure machine learning task continuously for 38 hours without human intervention, during which the AI independently diagnosed a 'shortcut learning bias' issue and launched 6 new experiments in parallel.66% similarUnverifiedThe training scale is 4096 games in parallel, about 330,000 agent steps per second end-to-end, completing about 19 billion agent steps in about 30 hours on a single Lambda GH200.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1152892API
curl https://kongchang.com/api/v1/knowledge/claims/1152892MCP
get_claim(id=1152892)