Compare Start here

xAILanguage modelSuperseded

Grok 4.6

Monthly-cadence Grok release that set up 4.7.

Compare head-to-head Race it in 3D

Overall Intelligence

#13

of 56
5.2 pts behind #1

0

races led

0

weeks at #1

83.5

score

Released: Aug 12, 2026; 2 mo ago
Context: 1M; 128K max output
Price: $2 / $6; per 1M in/out · $3.00 blended
Speed: 106 tok/s; 4s to first token
Modalities: text · image; outputs text
Weights: Closed; API and apps

The 30-second version

Monthly-cadence Grok release that set up 4.7.

Key takeaways

1Monthly-cadence Grok release that set up 4.7.

Where it ranks

Current snapshot · illustrative scores · click any race to watch it

Grok 4.6Best on each axis

Overall

13Overall+5.2 pts

Coding

14Coding+7.9 pts 14Frontend & UI+50 Elo 17Backend & APIs+8.4 pts 12Distributed Systems+8.4 pts 16Full-Stack Apps+9.4 pts 13Debugging & Repair+9.0 pts 15Agentic Coding+10.4 pts 14Code Review & Security+7.1 pts 11Data Science & SQL+4.1 pts 15Mobile Apps+7.0 pts 15Terminal & DevOps+11.2 pts 12Competitive Programming+118

Hard Reasoning

11Hard Reasoning+5.8 pts 10Frontier Questions+9.1 pts 9Abstract & Novel Reasoning+7.5 pts 11Logic & Puzzles+4.3 pts 17Planning & Strategy+8.4 pts

Math

10Math+4.9 pts 7Competition Math+4.0 pts 9Research Math+5.7 pts 10Applied & Quantitative+4.5 pts

Science

10Science+6.2 pts 15Graduate Science QA+4.7 pts 11Biology & Medicine+6.4 pts 10Physics & Chemistry+3.9 pts 10Research Assistance+5.7 pts

Writing & Language

15Writing & Language+7.4 pts 15Creative Writing+49 Elo 14Technical Writing+6.5 pts 17Editing & Style+6.6 pts 19Multilingual+5.7 pts 13Instruction Following+5.2 pts 16Conversation & EQ+47 Elo

Multimodal

15Multimodal+13.5 pts 22Image Understanding+10.7 pts 18Documents & Charts+9.0 pts 22Video Understanding+18.6 pts 9Speech & Audio+8.0 pts

Tool Use & Agents

17Tool Use & Agents+9.6 pts 21Tool Use+8.4 pts 19Computer Use+16.4 pts 12Web Research+8.1 pts 19Long-Horizon+20.4 hrs

Long Context

15Long Context+8.9 pts 191M Recall+8.2 pts 14Long Reasoning+6.8 pts

Speed & Cost

19Speed & Cost+34.8 pts 19Output Speed+417 tok/s 29Latency+3.70s 14Cost-Efficiency+34.5 pts

Trust & Factuality

16Trust & Factuality+9.6 pts 16Factuality+7.9 pts 16Honesty & Calibration+8.0 pts

Overall Intelligence over time

#1 frontierGrok 4.6Muse WonderGrok 4.7

Update history

The distilled version first; open any release for the deeper breakdown.

The Grok family

Grok 3February 2025 Grok 4July 2025 Grok 4.1November 2025 Grok 4.6August 2026 Grok 4.7September 2026

60-second briefing

Grok 4.6, explained

AI video summary · placeholder render from the auto-written script

From xAI's own guide, distilled

Prompt engineering for Grok

Hand Grok the relevant files and context up front instead of letting it guess.
State explicit goals and acceptance criteria; vague asks produce generic code.
Iterate quickly: Grok's speed makes several fast refinements cheaper than one long prompt.
Use agentic mode for multi-step work and keep the tool list focused.

All official guides, distilled →

About xAI

Builder of Grok, trained on the Colossus supercomputer and wired into X.

Optimising for

Raw reasoning and math from very large training runs, real-time knowledge from X and the web, and a fast release cadence.

Palo Alto, US · founded 2023

Closest rivals

Grok 4.6 is 13th in overall intelligence; leader is Claude Opus 5.5.