The Model Index
Up to four models

Claude Opus 4.6 vs Claude Opus 4.5

On the index (GPT-4 = 100), Claude Opus 4.6 scores 165 and Claude Opus 4.5 152. 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 Claude Opus 4.6
Presets
Claude Opus 4.6
165149–181
WeightsClosed
ReleasedFeb 2026
Price in / out$5.0 / $25
Context1M
Core tests taken7 of 9
Claude Opus 4.5
152136–168
WeightsClosed
ReleasedNov 2025
Price in / out$5.0 / $25
Context200K
Core tests taken5 of 9

Claude Opus 4.6 leads Claude Opus 4.5 by 13 points, ahead on 5 of 5 shared core tests, at a similar price.

Test by testthird-party results

Claude Opus 4.6Claude Opus 4.5
0%25%50%75%100%REASONING & KNOWLEDGEGPQA Diamond (Epoch AI)GPQA Diamond (Epoch AI)Humanity's Last ExamHumanity's Last ExamLiveBench Data AnalysisLiveBench Data AnalysisLiveBench IFLiveBench IFLiveBench LanguageLiveBench LanguageLiveBench ReasoningLiveBench ReasoningGPQA Diamond (Vals AI)GPQA Diamond (Vals AI)MMLU ProMMLU ProMMMU ProMMMU ProCL-benchCL-benchEBR-benchEBR-benchSimpleBenchSimpleBenchSimpleQA 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 CodingSWE-bench Pro (public set)SWE-bench Pro (public set)Terminal-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 BenchFurniture assemblyFurniture assemblyLMCALMCAMystery game puzzlesMystery game puzzlesPROFESSIONAL WORKCorpFin v2CorpFin v2LegalBenchLegalBenchMedCodeMedCodeMedScribeMedScribeMortgageTaxMortgageTaxTaxEval v2TaxEval v2
Every result as a table
TestClaude Opus 4.6Claude Opus 4.5
Reasoning & knowledge
GPQA Diamond (Epoch AI)90.5%86.1%
Humanity's Last Exam34.4%25.2%
LiveBench Data Analysis69.9%74.4%
LiveBench IF63.3%62.5%
LiveBench Language83.3%81.3%
LiveBench Reasoning88.7%80.1%
GPQA Diamond (Vals AI)89.6%85.9%
MMLU Pro89.1%87.3%
MMMU Pro83.9%83.0%
CL-bench20.7%21.1%
EBR-bench12.7%14.3%
SimpleBench67.6%62.0%
SimpleQA Verified47.0%45.7%
Maths
FrontierMath (tiers 1-3)66.0%34.4%
AIME-style maths (OTIS mock)94.4%86.1%
LiveBench Mathematics89.3%90.4%
Chess puzzles21.0%14.0%
FrontierMath (2025 set)40.7%20.7%
FrontierMath tier 4 (2025 set)22.9%4.2%
FrontierMath (tier 4)26.8%4.9%
Coding
SWE-bench Verified78.7%76.6%
LiveBench Agentic Coding49.0%39.7%
LiveBench Coding78.2%79.7%
SWE-bench Pro (public set)51.9%45.9%
Terminal-Bench 2.0 (Terminus 2 runs)62.9%57.8%
LiveCodeBench84.7%83.7%
SWE-bench78.2%76.4%
Vibe Code Bench v1.157.6%20.6%
GSO-Bench41.2%26.5%
WeirdML78.0%63.7%
Agents
DTBench91.2%89.9%
DeepResearch Bench55.3%54.8%
Furniture assembly28.3%28.3%
LMCA55.8%44.5%
Mystery game puzzles25.0%22.0%
Professional work
CorpFin v267.0%65.1%
LegalBench85.3%84.6%
MedCode49.1%49.2%
MedScribe86.7%85.3%
MortgageTax68.5%68.7%
TaxEval v276.0%74.9%
Task length · METR47 sec45 sec

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

This page as a chartDownload PNG