The Model Index
Up to four models

GPT-5.4 vs Grok 4.7

On the index (GPT-4 = 100), GPT-5.4 scores 166 and Grok 4.7 164. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.

Next to GPT-5.4
Presets
GPT-5.4
166150–182
WeightsClosed
ReleasedMar 2026
Price in / out$2.5 / $15
Context1.05M
Core tests taken7 of 9
Grok 4.7
164149–180
WeightsClosed
ReleasedSep 2026
Price in / out$2.0 / $6.0
Context500K
Core tests taken6 of 9

GPT-5.4 and Grok 4.7 are too close to call (1 point apart, within the uncertainty), ahead on 3 of 4 shared core tests, at 1.9× the price.

Test by testthird-party results

GPT-5.4Grok 4.7
0%25%50%75%100%REASONING & KNOWLEDGEGPQA Diamond (Epoch AI)GPQA Diamond (Epoch AI)SimpleQA VerifiedSimpleQA VerifiedMATHSFrontierMath (tiers 1-3)FrontierMath (tiers 1-3)AIME-style maths (OTIS mock)AIME-style maths (OTIS mock)Chess puzzlesChess puzzlesFrontierMath (tier 4)FrontierMath (tier 4)CODINGCode MigrationCode MigrationProgramBenchProgramBenchVibe Code Bench v1.1Vibe Code Bench v1.1MirrorCodeMirrorCodeAGENTSAPEX-AgentsAPEX-AgentsDTBenchDTBenchFurniture assemblyFurniture assemblyLMCALMCAMystery game puzzlesMystery game puzzlesPROFESSIONAL WORKLegalBenchLegalBenchMedCodeMedCodeMedScribeMedScribeNOVEL PROBLEMSARC-AGI-2ARC-AGI-2ARC-AGI-1ARC-AGI-1ARC-AGI-3ARC-AGI-3
Every result as a table
TestGPT-5.4Grok 4.7
Reasoning & knowledge
GPQA Diamond (Epoch AI)93.3%92.7%
SimpleQA Verified45.1%56.0%
Maths
FrontierMath (tiers 1-3)78.6%53.0%
AIME-style maths (OTIS mock)97.8%98.1%
Chess puzzles48.0%41.0%
FrontierMath (tier 4)49.0%17.1%
Coding
Code Migration35.0%44.8%
ProgramBench0.5%0.5%
Vibe Code Bench v1.167.4%86.2%
MirrorCode23.3%54.4%
Agents
APEX-Agents52.4%70.0%
DTBench94.4%96.0%
Furniture assembly37.5%20.8%
LMCA52.0%49.4%
Mystery game puzzles37.0%29.0%
Professional work
LegalBench86.0%84.4%
MedCode41.3%49.5%
MedScribe77.5%89.4%
Novel problems
ARC-AGI-274.0%61.4%
ARC-AGI-193.7%90.2%
ARC-AGI-30.2%1.8%
Task length · METR45 sec

Tests at least two of these models have taken. Bold rows are the core tests.

This page as a chartDownload PNG