Fable 5.2 excels at chip design and biology but Anthropic nerfed RSI tasks
omarsar0 · x · 2026-11-10
Citing Dylan Patel's eval post: Fable 5.2 is significantly better at chip design, software engineering, biology, and material science, per GDPx — the only eval testing these categories at this depth, with tasks written by true experts. Notably, it shows no gains in finance, legal, quant trading, RSI, or hardware design — Patel points out Anthropic explicitly nerfed the model on RSI tasks, reflected in the scores.
Elvis OmAr Sar (omarsar0) comments that this sucks for independent researchers, but credits Anthropic for being transparent about it this time. He argues open models are better suited for this kind of work — you can modify weights and get transparency — and says he's testing but hasn't found a solid open model with sparks of RSI yet; he hopes a strong open model will emerge to put pressure on labs.
More from Models
- ChatGPT/Codex desktop app's latest update breaks the send button, user reports — rudrank · 2026-11-10
- Fable 5.2 benchmarks, itemized: 90% of Anthropic code now Claude-written — anitakirkovska · 2026-11-10
- Claude Fable 5.2 beats trendline on all three unsaturated benchmarks — emollick · 2026-11-10
- Meta's Stealth Model Crouton Alpha Revealed as Early Muse Ember, 5T Tokens Processed — alexandr_wang · 2026-11-10
- OpenAI says GPT-6.1 Terra beats Sol on visual understanding; Claude Opus 5.5 tops chart — reach_vb · 2026-11-10
- Polymarket odds: Grok 5 by end of 2026 just 11%, by end of 2027 83% — Polymarket · 2026-11-10