AI news ·
Claude Fable 5.2 Completes 41.3% of Real Freelance Jobs in Benchmark
Claude Fable 5.2 completed 41.3% of real freelance tasks at an acceptable quality level, nearly double the next-best model.
A new benchmark report shows Claude Fable 5.2 completed 41.3% of real freelance projects in the Remote Labor Index at a quality level clients accepted. That rate is nearly twice the figure posted by the next-highest model in the same test set.
The result points to rapid improvement in multi-step knowledge work, where tasks require sustained reasoning, revision, and domain awareness rather than single-turn answers. For managers, creative directors, and research leads tracking operational readiness, the gap between the leading model and the rest of the field is now wide enough to affect pilot decisions.
Several factors still limit how directly the number translates into production settings. Cost per completed task, end-to-end speed, and how well the Remote Labor Index mirrors actual workflows all need further examination before the 41.3% figure can be treated as a reliable deployment forecast. The finding adds a concrete data point to discussions about task suitability, oversight requirements, and return on investment for teams evaluating AI assistance in writing, analysis, and project execution.