Fable 5 reaches 67.3% on GameDevBench, trailing only GPT-5.6 Sol
scaling01 · x · 2026-07-24
Fable 5 tops GameDevBench with 67.3% task success, behind only GPT-5.6 Sol
A shared benchmark graphic says Fable 5 is now the best model on GameDevBench among the listed entries, with a 67.3% task success rate.
The same chart shows it:
- Beats every model except GPT-5.6 Sol
- Sits above several GPT-5.6 Sol variants and Claude Opus 4-8
- Represents a big jump from a few months ago, when models reportedly could not even break 50%
More from coding & agent
- A technical AI crash course covers LLMs, MCP, agents, skills, and RAG — aakashgupta · 2026-07-24
- The Stack v3 becomes the largest open code dataset at 114 TB and 5T tokens — JJitsev · 2026-07-24
- Cursor launches an intelligent model router for coding requests, claiming 60% lower cost — Arindam_1729 · 2026-07-24
- Opinion: AI Agent Harnesses Will Evolve From Products to Libraries — samgoodwin89 · 2026-07-24
- An engineering lead uses OpenLoomi to sync GitHub, Linear, and PR reviews — Yuuyake · 2026-07-24
- Gemini CLI patches an infinite OAuth loop by serializing credential writes — EngKMM · 2026-07-24