Claude Fable 5.2 just topped the Artificial Analysis Intelligence Index at 65 — has anyone actually felt the jump over Opus 5.5?
retained text
Anthropic released Fable 5.2 and Mythos 5.2 on Monday. Fable is the public version at $8/$40 per million tokens, while Mythos is the same underlying model restricted to vetted cybersecurity and life-science organizations. Anthropic claims 69.9% on SWE v2 against 62.3% for Opus 5.5, and the independent Artificial Analysis index now ranks Fable first, ahead of Gemini 4 Carbon. For those who have tried it: is the difference obvious in real coding or research work, or is this another benchmark jump you only feel on the hardest tasks? And what do you think of the two-tier release, one model for everyone and a restricted twin for vetted organizations only?
document fetched 2026-11-14 00:22 · canonical https://reddit.com/r/ClaudeAI/comments/1xvbe0t/claude_fable_52_just_topped_the_artificial