Unverified50% confidenceBenchmarkExact time
On the Agent Tech scientific research benchmark, Fable 5.2 scored 74.6%, Opus 5.5 scored 58.7%, Astra scored 68.1%, and Carbon scored 64.8%.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
11/12/2026
First Seen
Valid until: 11/26/2026
Sources
Anthropic Releases Fable 5.2: Billed as the World's Most Powerful AI Model
bilibiliAI译片君11/11/2026
Related Entities
Related Claims
UnverifiedIn the official Oswald test, Astra scored 72.6%, while the previous generation Cell scored 65.7%.70% similarUnverifiedIn the system test simulating a human operating a computer, Claude Opus 5.5 scored 87.2% and Carbon scored 81.5%.69% similarUnverifiedGPT-6 Astra achieved a near-perfect score of 99.9% on ARC-AGI-type tests, while the average score of ordinary human test-takers was only 48%.64% similarUnverifiedAn ECE between 3.7 and 5.5 means that when the model gives 80% confidence, the actual accuracy is roughly between 74.5% and 76.3%.64% similarUnverifiedIn a test checking whether AI-modified code could be accepted by a project, Opus 5.5 using the default setting scored slightly higher than Astra's best publicly available score, and the cost per task was approximately one-fifth of Astra's.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1163379API
curl https://kongchang.com/api/v1/knowledge/claims/1163379MCP
get_claim(id=1163379)