Claude Fable 5 tops MirrorCode with a 64% solve rate, far ahead of GPT-5.6 Sol
davidad · x · 2026-08-04
Epoch AI updated the MirrorCode leaderboard with Claude Fable 5 and GPT-5.6 Sol.
- Claude Fable 5 now leads with a 64% solve rate.
- GPT-5.6 Sol follows at 20%.
- The post frames this as a major software-engineering advantage for Fable.
Related event: Claude Fable 5 Tops EpochAI's Updated MirrorCode Leaderboard(4 posts)→
More from Models
- Qwen3-VL-32B variant starts trending on Hugging Face — ethanfel · 2026-08-04
- Dev Vows to Turn US AI Bear If DeepSeek Nails Complex Task — yacineMTB · 2026-08-04
- DeepSeek Flash takes a reverse-engineering task ChatGPT refused — yacineMTB · 2026-08-04
- A model that ‘won 10 Fields Medals’ still recommended a cafe closed two years ago — bosmeny · 2026-08-04
- Open models still have one big advantage: you can read the reasoning thread — jamesdouma · 2026-08-04
- opencode’s model picker now includes a Canadian model — yacineMTB · 2026-08-04