Grok 4.7 scores 46 on AA Intelligence Index, enters top 4 labs but burns 81k tokens per task
ArtificialAnlys · x · 2026-09-22
Artificial Analysis evaluated Grok 4.7, bringing xAI into the top 4 labs:
- Intelligence Index: 46, +2 over Grok 4.6 (evaluated at xhigh reasoning effort)
- Agentic knowledge work: 1657 Elo on AA-Briefcase (+111), joining Claude Opus 5 and Claude Fable 5.1 at the frontier; 1695 Elo on GDPval-AA
- Coding agents: 56 on the Coding Agent Index with Grok Build (+9), ranking 4th natively, overtaking GPT-5.6 Sol
- Token cost: 81k output tokens per task, 125% more than Grok 4.6 (36k) and 196% more than GPT-6 Astra (27k)
- Hallucination rate improved 34%→29%; Terminal-Bench 4.0 +4.5pp, AA-LCR -3.7pp
- 500k context unchanged; pricing $2/$6 per 1M tokens ($0.50 cache hits); 188 tokens/sec output
More from Models
- Agent Arena publishes a cost-performance Pareto view comparing agents — arena · 2026-11-12
- Upgrading to Google AI Ultra broke Gemini's Chrome browsing that worked on Pro — batmanButthole · 2026-11-12
- Claude Fable 5.2 tops Code4Scene at 84.4/100 — and gamed the collision checker — ZhitingHu · 2026-11-12
- Google engineer spotlights AI Studio redesign, name-drops unreleased Gemini 4 Carbon — ammaar · 2026-11-12
- All frontier models cheat on HoneyBench v0.2, with cheat rates from 12% to 86% — ChowdhuryNeil · 2026-11-12
- Google DeepMind Code Summit talk urges devs to prepare for massive inference speed gains — altryne · 2026-11-12