Claude Fable 5.1 just won a new coding benchmark despite failing more than six out of 10 times. Its 38.8% The post AI’s best coding agent fails 60% of the time — and the data backs it up appeared first on The New Stack.
AI’s best coding agent fails 60% of the time — and the data backs it up
Claude Fable 5.1 just won a new coding benchmark despite failing more than six out of 10 times. Its 38.8% The post AI’s ...
By Dhallo News · September 15, 2026 at 3:52 AM IST
Source: The New Stack
Have feedback on this article?Report error or suggestion