Claude Fable 5.2 benchmark dump: sweeps 12 evals, robotics jumps to 64.9%
mark_k · x · 2026-11-10
markk shared the official Claude Fable 5.2 benchmark table vs Fable 5.1, Opus 5.5, GPT-6 Astra, and Gemini 4 Carbon. Highlights:
- Agentic coding: FrontierCode 2.0 49.1% (vs 32.4% for 5.1), FrontierSWE v2 69.9%
- Scientific research 74.6%, cybersecurity 76.3%/89.6% (Mythos 5.2)
- Protein design 85.1%, OSWorld 2.1 computer use 89.3%, robotics 64.9% (up from 27.6%)
- Humanity's Last Exam with tools 74.2%; AA-Briefcase Elo 1785 leads, but AutomationBench 51.4% trails Gemini 4 Carbon's 53.9%
Some rows are scored by Mythos 5.2; some Opus numbers are noted as without safeguards.
Related event: Claude Fable 5.2 Launches, Leading Opus 5.5 and GPT-6 Across Benchmarks(5 posts)→
More from Models
- Claude Fable 5.2 tops FrontierCode 2.0 at 49.1, at a third of Fable 5.1's cost — nrehiew_ · 2026-11-10
- Polymarket puts 52% odds on next Claude Opus landing this month as Anthropic ships 5.2 models — Polymarket · 2026-11-10
- Artificial Analysis index: Claude Fable 5.2 tops Gemini 4 Carbon, 65 vs 60 — Angaisb_ · 2026-11-10
- Anthropic: Claude Fable 5.2 designs working electronic hardware end to end — EricBuess · 2026-11-10
- Claude Mythos 5.2 redisCOVERS 89.6% of known CVEs in Anthropic's internal eval — minchoi · 2026-11-10
- Claude Mythos 5.2 accurately models 28 of 34 viral capsids for gene therapy research — minchoi · 2026-11-10