Top 15 of 35 scored · 21 benchmarks · method v1
Overall: 1. Claude Fable 5.1, 2. Claude Opus 5, 3. GPT-6 Astra
Leader
92
Claude Fable 5.1 · Anthropic
Best open weights
64
Kimi K3 · Moonshot
Best value
53
GLM-5.3-Flash · Z.ai (Zhipu AI)$0.50 per 1M output tokens