Name: Grok 4.7
Author: xAI

Grok 4.7

xAI · 2026-09-21

Grok 4.7 is SpaceXAI's September 2026 model for coding, agentic tasks, and knowledge work. It supports low, medium, high, and xhigh reasoning effort, with high as the default. Grok 4.7 Fast serves the same model on faster infrastructure. Its published knowledge cutoff is May 2026, without day precision.

Benchmark scores

Benchmark	Score
AA-Briefcase v1.1 Elo	1657
AA-LCR v1.1	77.0
AA-Omniscience Index	32.0
Artificial Analysis Intelligence Index v4.3.2	46.4
Artificial Analysis Omniscience Accuracy	47.8
Artificial Analysis Omniscience Hallucination Rate	29.3
AutomationBench-AA 1.0.6	65.6
Capability	162.1
CritPt (avg@5)	18.0
FrontierCode 1.1 Extended - Pass Rate	65.2
FrontierCode 1.1 Extended - Score	59.4
FrontierCode 1.1 Main - Pass Rate	53.1
FrontierCode 1.1 Main - Score	47.6
GDP.pdf (Artificial Analysis, All-pass)	23.2
GDP.pdf (Artificial Analysis, Mean Criteria)	71.2
GDPval-AA v2.1	1695
Humanity's Last Exam (Text-Only)	43.1
Medical Long Context Reasoning (MLCR-AA)	15.0
MLCR-AA Accuracy (Judged Responses)	84.7
MLCR-AA Completeness (Judged Responses)	18.8
MLCR-AA Conciseness	97.8
Next.js AI Agent Evaluations	94.0
SciCode 1.0.1 (Artificial Analysis)	57.8
Terminal-Bench 4.0 (Artificial Analysis)	25.8