Claude Fable 5.2 Tops Intelligence Index at 65, Price Cut 20% but Cost per Task 2.6x Opus 5.5
ArtificialAnlys · x · 2026-11-10
Artificial Analysis benchmarks Anthropic's Claude Fable 5.2, which takes #1 on Intelligence Index v5 at 65, ahead of Gemini 4 Carbon (60) and Claude Opus 5.5 (59), up 15 points from Fable 5.1.
Highlights
- Leads 5 of 9 evals: HLE-Diamond 80% (11 pts over Gemini 4 Carbon), AA-Omniscience v1.1 67, Terminal-Bench-Science 73%, CritPt v2 hard 77%, FrontierCode 2.0 49%
- Gains driven by scientific reasoning, knowledge and coding; agentic knowledge work (AA-Briefcase 1785 Elo) and Terminal-Bench 4.0 (60%) only roughly match Opus 5.5
- Priced $8/$40 per 1M tokens (20% cheaper than Fable 5.1); 1M context, five effort settings
Cost caveat
- Max effort costs $26.5/task, 2.6x Opus 5.5 ($10.0)
- But high/xhigh settings score 62 — beating every model at any effort — with 40% fewer output tokens than Opus 5.5 max; low effort beats Fable 5.1 max with 70% fewer tokens
Related event: Fable 5.2 tops intelligence index but per-task cost draws skepticism(3 posts)→
More from Models
- Claude Fable 5.2 lands on Databricks Day 0, tops OfficeQA Pro v2 benchmark — matei_zaharia · 2026-11-10
- Can you nerf a model without hurting mech interp? Researchers weigh in — eliebakouch · 2026-11-10
- Users ask if Claude Opus 5.5 has already been quietly nerfed — eigenron · 2026-11-10
- Ant Group's Robbyant open-sources LingBot Touch, a JEPA-based tactile foundation model — AdinaYakup · 2026-11-10
- Fable 5.2 not yet showing up in Claude Code for some users — DavidmComfort · 2026-11-10
- Claude Fable 5.2 designs 3 working electronic devices from one-paragraph prompts — Dr_Singularity · 2026-11-10