Grok 4.7 joins top four on Artificial Analysis
Grok 4.7 scores 46 on Artificial Analysis's Intelligence Index, putting SpaceXAI among the top four labs, with large agentic and coding gains that come at more than double the token usage.
Artificial Analysis has published independent benchmarks for Grok 4.7, evaluated at xhigh reasoning effort. The model scores 46 on the firm's Intelligence Index, 2 points above Grok 4.6, which places SpaceXAI among the top four AI labs.
The biggest gains are in agentic knowledge work. On AA-Briefcase, Artificial Analysis's private benchmark for long-horizon professional tasks, Grok 4.7 scores 1657 Elo. That is up 111 from Grok 4.6 (high) and just behind Claude Opus 5 and Claude Fable 5.1. Most of the improvement is in analytical quality (1994 Elo, up from 1690), while presentation quality slipped slightly, from 1519 to 1499. On GDPval-AA, it scores 1695 Elo, up from 1605.
In xAI's own Grok Build harness, Grok 4.7 scores 56 on the Coding Agent Index, up from 47. That ranks it 4th among models running in their native harnesses, behind Claude Fable 5.1, GPT-6 Astra and Claude Opus 5, and ahead of GPT-5.6 Sol. It improved on all three components: DeepSWE v1.1 rose from 65% to 73%, Terminal-Bench 4.0 from 18% to 33%, and SWE-Atlas-QnA from 58% to 63%.
On the other Intelligence Index tasks, results are broadly flat. AA-LCR fell 3.7 points and AutomationBench-AA fell 1.1. The hallucination rate dropped from 34% to 29%.
The gains cost more tokens. Grok 4.7 uses about 81k output tokens per Index task, compared with 38k for Grok 4.6 (xhigh) and 27k for GPT-6 Astra (max). It averaged about 7.1 minutes per task. Pricing is unchanged at $2/$6 per million input/output tokens, and the context window remains 500k tokens.