Claude Fable Sets High Score on BU Bench

Claude Fable 5 achieved an 80% score on BU Bench V1, outperforming competing models on web task completion, though at a higher API cost of $580.87 for 100 tasks. Testing showed Fable excels at maintaining task continuity and reasoning through complex web interactions.