Fable 5 Fails All Hardest Tasks in UC Berkeley ALE Exam, Costs 4-12x More Than Rivals

According to UC Berkeley RDI, the latest Agents' Last Exam (ALE) evaluation results released this week show a 0% success rate on the hardest tasks requiring sustained reasoning and deep expertise across all tested AI agents, including newly released Fable 5. In per-task API costs, Fable 5 charged $15.70—4 times higher than GPT-5.5 at $3.80 and 12 times higher than Composer 2.5 at $1.33. The evaluation covered 55 professional domains with over 1,500 expert-verified tasks and found that agents most commonly fail by prematurely declaring success without validating results.
Disclaimer: The information on this page may come from third-party sources and is for reference only. It does not represent the views or opinions of Gate and does not constitute any financial, investment, or legal advice. Virtual asset trading involves high risk. Please do not rely solely on the information on this page when making decisions. For details, see the Disclaimer.
Comment
0/400
No comments