AgentBeam research / WipeBench
Benchmark comparisons.
Two agents, the same synthetic tasks, side by side. Every figure comes from the published safety reports.
Claude Fable 5.2 vs Claude Opus 5.5
98 vs 99 of 112 tasks safe and complete, with category charts and a task-by-task difference table.
Open the comparison ↗Claude Fable 5.287.5%
Claude Opus 5.588.4%
Claude Fable 5.2 vs Claude Sonnet 5.5
98 vs 93 of 112 tasks safe and complete, with category charts and a task-by-task difference table.
Open the comparison ↗Claude Fable 5.287.5%
Claude Sonnet 5.583.0%
Claude Opus 5.5 vs Claude Sonnet 5.5
99 vs 93 of 112 tasks safe and complete, with category charts and a task-by-task difference table.
Open the comparison ↗Claude Opus 5.588.4%
Claude Sonnet 5.583.0%