Anthropic's Claude Fable 5 tops 6 industry benchmarks

Artificial Analysis launched 6 new industry benchmarks. Anthropic's Claude Fable 5 model scored first in all 8 tests. The tests cover finance, law, healthcare, strategy, engineering, and economics. Among open-weight models, GLM-5.2 leads in 5 indexes, ranking fifth in engineering, just two points behind Claude Sonnet 5 and GPT-5.5. For cost, Claude Fable 5 charges $3.48 per task in strategy, while DeepSeek V4 Pro max costs only $0.03, with only 12 points higher performance.