Model2h ago
Claude Opus 5 Matches Fable 5 Performance
Anthropic released Claude Opus 5, which matches Fable 5 performance while cutting token costs by 50%. It scored 30.2 percent on the ARC-AGI-3 benchmark, nearly quadrupling the result of GPT-5.6 Sol. These gains in coding and knowledge work lower the barrier for high-reasoning tasks. Developers now have a cheaper alternative for complex problem-solving.