Gemini 3.1 Pro Preview vs Grok 4.6

Google against xAI. Prices, scores and speed below are live.

  • Grok 4.6 costs 33% less on blended price ($4.50 against $6.75 per 1M tokens, 3 input to 1 output).
  • Grok 4.6 scores higher on the Artificial Analysis Intelligence Index: 31 against 30.4.
  • On the 47 benchmarks both have scores on, Gemini 3.1 Pro Preview leads on 18 and Grok 4.6 on 29.
  • Gemini 3.1 Pro Preview reads more context: 1M tokens against 500K.
  • Grok 4.6 is newer: listed Aug 12, 2026, against Feb 19, 2026.

Side by side

Gemini 3.1 Pro PreviewGrok 4.6
LabGooglexAI
ListedFeb 19, 2026Aug 12, 2026
Input per 1M tokens$3.00$3.00
Output per 1M tokens$18.00$9.00
Cached input per 1M$0.30$0.75
Blended (3:1) per 1M$6.75$4.50
Context window1M500K
Max output66K450K
Knowledge cutoff--
Intelligence index30.431
Coding index68.875.9
ReasoningYesYes
Reads imagesYesYes
Tool useYesYes
Structured outputYesYes

Every benchmark both are scored on

Each model's best score per benchmark, with its source. Scores from different sources or settings are not always comparable; each links to where it was published.

BenchmarkGemini 3.1 Pro PreviewGrok 4.6
AIME95.6%Epoch AI99.2%Epoch AI
APEX-Agents35.1%Epoch AI65.3%Epoch AI
ARC-AGI-198%Epoch AI87.5%Epoch AI
ARC-AGI-277.1%Epoch AI67.1%Epoch AI
Artificial Analysis Agentic Index8.2Artificial Analysis and OpenRouter53Artificial Analysis and OpenRouter
Artificial Analysis Coding Index68.8Artificial Analysis and OpenRouter76.8Artificial Analysis and OpenRouter
Artificial Analysis Intelligence Index30.4Artificial Analysis and OpenRouter44.4Artificial Analysis and OpenRouter
Chess Puzzles51%Epoch AI33%Epoch AI
DeepSWE11.7%Epoch AI67.5%Epoch AI
Design Arena: 3D1231 EloDesign Arena1273 EloDesign Arena
Design Arena: Android Native (agentic)1045 EloDesign Arena1294 EloDesign Arena
Design Arena: ASCII Art1267 EloDesign Arena1272 EloDesign Arena
Design Arena: Code1242 EloDesign Arena1290 EloDesign Arena
Design Arena: Data Viz1239 EloDesign Arena1283 EloDesign Arena
Design Arena: Full-stack (agentic)1068 EloDesign Arena1264 EloDesign Arena
Design Arena: Game Dev1202 EloDesign Arena1295 EloDesign Arena
Design Arena: Game Dev (agentic)1061 EloDesign Arena1158 EloDesign Arena
Design Arena: Godot Game Dev (agentic)1236 EloDesign Arena1276 EloDesign Arena
Design Arena: HTML Slides (agentic)1049 EloDesign Arena1146 EloDesign Arena
Design Arena: Mobile Apps (agentic)1093 EloDesign Arena1232 EloDesign Arena
Design Arena: PPTX Slides (agentic)1109 EloDesign Arena1197 EloDesign Arena
Design Arena: SVG1281 EloDesign Arena1234 EloDesign Arena
Design Arena: UI Component1275 EloDesign Arena1290 EloDesign Arena
Design Arena: Web Apps (agentic)1113 EloDesign Arena1235 EloDesign Arena
Design Arena: Website1251 EloDesign Arena1287 EloDesign Arena
DTBench97.1%Epoch AI97.3%Epoch AI
Earthborne Rangers (EBR-bench)14.3Epoch AI30.5Epoch AI
FrontierMath Tier 4 (v2)26.8%Epoch AI31.7%Epoch AI
FrontierMath Tiers 1-3 (v2)59.6%Epoch AI66%Epoch AI
GPQA Diamond94.5%Artificial Analysis and OpenRouter92.8%Artificial Analysis and OpenRouter
LMArena (Coding)1484 EloLMArena1464 EloLMArena
LMArena (Overall)1480 EloLMArena1429 EloLMArena
LMArena: Creative Writing1482 EloLMArena1426 EloLMArena
LMArena: Hard Prompts1485 EloLMArena1447 EloLMArena
LMArena: Longer Query1482 EloLMArena1456 EloLMArena
LMArena: Math1478 EloLMArena1461 EloLMArena
LMArena: Multi-Turn1486 EloLMArena1425 EloLMArena
LMCA53.8Epoch AI48.5Epoch AI
MirrorCode (ML, +Private, 2L)8.9Epoch AI41.1Epoch AI
Mystery Game Puzzles32%Epoch AI34%Epoch AI
ProofBench26%Epoch AI51%Epoch AI
SciCode58.7%Epoch AI56.5%Epoch AI
SimpleBench79.6%Epoch AI75.9%Epoch AI
SimpleQA Verified73.5%Epoch AI49.3%Epoch AI
tau-bench (Airline)77%Artificial Analysis and OpenRouter75.3%Artificial Analysis and OpenRouter
WebDev Arena1447 EloEpoch AI1618 EloEpoch AI
WeirdML72.1%Epoch AI67.3%Epoch AI

More comparisons

Where these figures come from

Prices are GenMagic's own, what the studio and the API charge, recomputed from each provider's price list every few minutes. The indices are the Artificial Analysis indices, and every benchmark score links to its source, re-read daily. Measured speed is GenMagic's own generation records, in aggregate. See Google models, xAI models, every LLM's price and our sources.