Artificial Analysis just shipped Intelligence Index v5, and the ranking is the… · @harden_369741 · TwKit

@harden_369741

Artificial Analysis just shipped Intelligence Index v5, and the ranking is the least interesting part. Claude Opus 5.5 leads at 58, GPT-6 Astra at 57, GPT-6.1 Sol at 55. A one point gap is noise, I would not switch models over it. What actually changed my mind is the plumbing. For the first time, 60% of the index runs on private test sets or held-out answers, and the open weights gap widened from 3 points to 12. Public leaderboard gaming just got harder, which is good, and the open side falling further behind at the top is not. The number I keep staring at is cost per task: $0.96 for GPT-6.1 Terra vs $13.04 for Claude Fable 5.1 among the top models. With that 13x spread, which number should decide a real deployment, the rank or the cost per task? Source: Artificial Analysis Intelligence Index v5 announcement. #AI #LLM

曝光 28 · 评论 0 · 点赞 0 · 书签 0