Anthropic's Fable 5.2 completes 41 percent of real freelance tasks
On the Remote Labor Index, Anthropic's AI system Fable 5.2 achieves a completion rate of 41.3 percent on real freelance tasks, more than double the rate of the next-best systems.
Anthropic's AI system Fable 5.2 completed 41.3 percent of the tasks in a test using real freelance jobs. The results come from the Remote Labor Index, a benchmark based on actual freelance work rather than synthetic test tasks. The quality of the delivered work was rated as professional.
The gap compared with other systems is substantial. According to the reporting, the next-best models, Opus 5.5 and GPT 6 Astra, reach less than half of Fable 5.2's completion rate. A year ago, the highest score in the same test was only 2.5 percent. The jump within one year is therefore considerable.
The benchmark also shows limits. In product design and animation, the system continues to perform weakly. In particular for creative design tasks, its performance lags behind that for other job types.
The figures come from reporting by The Decoder and were not independently verified for this article. They describe performance on the Remote Labor Index and do not permit general conclusions about the freelance labor market.
Source: the-decoder.com
4
