The Model Index
Up to four models

GPT-5.5 vs GPT-5.4

On the index (GPT-4 = 100), GPT-5.5 scores 172 and GPT-5.4 166. 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-5.5
Presets
GPT-5.5
172156–188
WeightsClosed
ReleasedApr 2026
Price in / out$5.0 / $30
Context1.05M
Core tests taken6 of 9
GPT-5.4
166150–182
WeightsClosed
ReleasedMar 2026
Price in / out$2.5 / $15
Context1.05M
Core tests taken7 of 9

GPT-5.5 leads GPT-5.4 by 6 points, ahead on 5 of 5 shared core tests, at 2.0× the price.

Test by testthird-party results

GPT-5.5GPT-5.4
0%25%50%75%100%REASONING & KNOWLEDGEGPQA Diamond (Epoch AI)GPQA Diamond (Epoch AI)LiveBench Data AnalysisLiveBench Data AnalysisLiveBench IFLiveBench IFLiveBench LanguageLiveBench LanguageLiveBench ReasoningLiveBench ReasoningGPQA Diamond (Vals AI)GPQA Diamond (Vals AI)MMLU ProMMLU ProMMMU ProMMMU ProEBR-benchEBR-benchSimpleQA VerifiedSimpleQA VerifiedMATHSFrontierMath (tiers 1-3)FrontierMath (tiers 1-3)AIME-style maths (OTIS mock)AIME-style maths (OTIS mock)LiveBench MathematicsLiveBench MathematicsChess puzzlesChess puzzlesFrontierMath (2025 set)FrontierMath (2025 set)FrontierMath tier 4 (2025 set)FrontierMath tier 4 (2025 set)FrontierMath (tier 4)FrontierMath (tier 4)CODINGSWE-bench VerifiedSWE-bench VerifiedLiveBench Agentic CodingLiveBench Agentic CodingLiveBench CodingLiveBench CodingCode MigrationCode MigrationLiveCodeBenchLiveCodeBenchProgramBenchProgramBenchSWE-benchSWE-benchVibe Code Bench v1.1Vibe Code Bench v1.1DeepSWEDeepSWEGSO-BenchGSO-BenchMirrorCodeMirrorCodeWeirdMLWeirdMLAGENTSAPEX-AgentsAPEX-AgentsSkillsBenchSkillsBenchDTBenchDTBenchDeepResearch BenchDeepResearch BenchFurniture assemblyFurniture assemblyLMCALMCAMystery game puzzlesMystery game puzzlesPROFESSIONAL WORKCorpFin v2CorpFin v2LegalBenchLegalBenchMedCodeMedCodeMedScribeMedScribeMortgageTaxMortgageTaxTaxEval v2TaxEval v2NOVEL PROBLEMSARC-AGI-2ARC-AGI-2ARC-AGI-1ARC-AGI-1ARC-AGI-3ARC-AGI-3
Every result as a table
TestGPT-5.5GPT-5.4
Reasoning & knowledge
GPQA Diamond (Epoch AI)94.0%93.3%
LiveBench Data Analysis81.6%79.3%
LiveBench IF70.7%70.2%
LiveBench Language87.4%82.6%
LiveBench Reasoning89.6%88.1%
GPQA Diamond (Vals AI)93.2%91.7%
MMLU Pro88.1%87.5%
MMMU Pro88.3%87.5%
EBR-bench34.3%25.4%
SimpleQA Verified63.0%45.1%
Maths
FrontierMath (tiers 1-3)85.3%78.6%
AIME-style maths (OTIS mock)100.0%97.8%
LiveBench Mathematics95.9%94.2%
Chess puzzles63.0%48.0%
FrontierMath (2025 set)51.7%47.6%
FrontierMath tier 4 (2025 set)35.4%27.1%
FrontierMath (tier 4)72.5%49.0%
Coding
SWE-bench Verified80.6%76.9%
LiveBench Agentic Coding54.0%53.8%
LiveBench Coding82.2%77.5%
Code Migration45.2%35.0%
LiveCodeBench85.3%84.1%
ProgramBench0.5%0.5%
SWE-bench82.6%78.2%
Vibe Code Bench v1.169.8%67.4%
DeepSWE67.0%51.8%
GSO-Bench40.2%31.4%
MirrorCode23.3%23.3%
WeirdML84.9%77.7%
Agents
APEX-Agents55.1%52.4%
SkillsBench62.2%51.7%
DTBench96.0%94.4%
DeepResearch Bench54.0%35.1%
Furniture assembly44.2%37.5%
LMCA54.3%52.0%
Mystery game puzzles56.0%37.0%
Professional work
CorpFin v268.4%65.3%
LegalBench86.5%86.0%
MedCode49.1%41.3%
MedScribe86.9%77.5%
MortgageTax68.8%68.3%
TaxEval v275.0%74.0%
Novel problems
ARC-AGI-285.0%74.0%
ARC-AGI-195.0%93.7%
ARC-AGI-30.4%0.2%
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