The Model Index
Up to four models

GPT-5.2 vs GPT-5.1

On the index (GPT-4 = 100), GPT-5.2 scores 162 and GPT-5.1 150. 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-5.2
Presets
GPT-5.2
162146–178
WeightsClosed
ReleasedDec 2025
Price in / out$1.75 / $14
Context400K
Core tests taken6 of 9
GPT-5.1
150134–166
WeightsClosed
ReleasedNov 2025
Price in / out$1.25 / $10
Context400K
Core tests taken4 of 9

GPT-5.2 leads GPT-5.1 by 12 points, ahead on 4 of 4 shared core tests, at 1.4× the price.

Test by testthird-party results

GPT-5.2GPT-5.1
0%25%50%75%100%REASONING & KNOWLEDGEGPQA Diamond (Epoch AI)GPQA Diamond (Epoch AI)Humanity's Last ExamHumanity's Last ExamGPQA Diamond (Vals AI)GPQA Diamond (Vals AI)MMLU ProMMLU ProMMMU ProMMMU ProCL-benchCL-benchSimpleBenchSimpleBenchSimpleQA VerifiedSimpleQA VerifiedVPCTVPCTMATHSAIME-style maths (OTIS mock)AIME-style maths (OTIS mock)MGSMMGSMChess puzzlesChess puzzlesFrontierMath (2025 set)FrontierMath (2025 set)FrontierMath tier 4 (2025 set)FrontierMath tier 4 (2025 set)CODINGSWE-bench VerifiedSWE-bench VerifiedTerminal-Bench 2.0 (Terminus 2 runs)Terminal-Bench 2.0 (Terminus 2 runs)LiveCodeBenchLiveCodeBenchSWE-benchSWE-benchVibe Code Bench v1.1Vibe Code Bench v1.1GSO-BenchGSO-BenchWeirdMLWeirdMLAGENTSDTBenchDTBenchDeepResearch BenchDeepResearch BenchLMCALMCAMystery game puzzlesMystery game puzzlesPROFESSIONAL WORKCorpFin v2CorpFin v2LegalBenchLegalBenchMedCodeMedCodeMedScribeMedScribeMortgageTaxMortgageTaxTaxEval v2TaxEval v2NOVEL PROBLEMSARC-AGI-2ARC-AGI-2ARC-AGI-1ARC-AGI-1
Every result as a table
TestGPT-5.2GPT-5.1
Reasoning & knowledge
GPQA Diamond (Epoch AI)91.4%87.6%
Humanity's Last Exam27.8%23.7%
GPQA Diamond (Vals AI)91.7%86.6%
MMLU Pro86.2%86.4%
MMMU Pro86.7%83.2%
CL-bench18.2%23.7%
SimpleBench45.8%53.2%
SimpleQA Verified37.1%48.0%
VPCT84.0%58.7%
Maths
AIME-style maths (OTIS mock)96.1%88.6%
MGSM94.0%93.0%
Chess puzzles48.0%42.0%
FrontierMath (2025 set)40.7%31.0%
FrontierMath tier 4 (2025 set)18.8%12.5%
Coding
SWE-bench Verified73.8%68.0%
Terminal-Bench 2.0 (Terminus 2 runs)54.0%47.6%
LiveCodeBench85.4%86.5%
SWE-bench75.8%69.8%
Vibe Code Bench v1.153.5%24.6%
GSO-Bench27.5%13.7%
WeirdML72.2%60.8%
Agents
DTBench90.9%90.1%
DeepResearch Bench41.1%42.8%
LMCA43.9%43.9%
Mystery game puzzles23.0%19.0%
Professional work
CorpFin v265.9%63.8%
LegalBench82.8%85.7%
MedCode49.8%52.7%
MedScribe84.4%88.1%
MortgageTax67.1%61.4%
TaxEval v275.8%74.9%
Novel problems
ARC-AGI-272.9%17.6%
ARC-AGI-194.5%72.8%
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