The Model Index
Up to four models

DeepSeek-V4-Pro vs GPT-5.6 Luna

On the index (GPT-4 = 100), DeepSeek-V4-Pro scores 163 and GPT-5.6 Luna 162. 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 DeepSeek-V4-Pro
Presets
DeepSeek-V4-Pro
163147–178
WeightsOpen
ReleasedAug 2026
Price in / out$0.11 / $4.44
Context1.05M
Core tests taken6 of 9
GPT-5.6 Luna
162146–178
WeightsClosed
ReleasedJul 2026
Price in / out$0.20 / $1.2
Context1.05M
Core tests taken6 of 9

DeepSeek-V4-Pro and GPT-5.6 Luna are too close to call (0 points apart, within the uncertainty), ahead on 4 of 6 shared core tests, at 2.7× the price.

Test by testthird-party results

DeepSeek-V4-ProGPT-5.6 Luna
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 ProSimpleQA VerifiedSimpleQA VerifiedMATHSFrontierMath (tiers 1-3)FrontierMath (tiers 1-3)AIME-style maths (OTIS mock)AIME-style maths (OTIS mock)LiveBench MathematicsLiveBench MathematicsProofBench v1.1ProofBench v1.1Chess puzzlesChess puzzlesFrontierMath (tier 4)FrontierMath (tier 4)CODINGTerminal-Bench 4.0Terminal-Bench 4.0LiveBench Agentic CodingLiveBench Agentic CodingLiveBench CodingLiveBench CodingCode MigrationCode MigrationCyberBench v1.1CyberBench v1.1IOIIOIProgramBenchProgramBenchSWE-benchSWE-benchVibe Code Bench 1-100Vibe Code Bench 1-100Vibe Code Bench v1.1Vibe Code Bench v1.1MirrorCodeMirrorCodeWeirdMLWeirdMLAGENTSAPEX-AgentsAPEX-AgentsSkillsBenchSkillsBenchSRE BenchSRE BenchTerminal-Bench-ScienceTerminal-Bench-ScienceDTBenchDTBenchLMCALMCAMystery game puzzlesMystery game puzzlesPROFESSIONAL WORKFinance Agent (v2)Finance Agent (v2)CorpFin v2CorpFin v2EMBEMBLegalBenchLegalBenchLegal Research BenchLegal Research BenchMedCodeMedCodeMedScribeMedScribeTax Agent BenchTax Agent BenchTaxEval v2TaxEval v2NOVEL PROBLEMSARC-AGI-2ARC-AGI-2ARC-AGI-1ARC-AGI-1
Every result as a table
TestDeepSeek-V4-ProGPT-5.6 Luna
Reasoning & knowledge
GPQA Diamond (Epoch AI)91.7%91.6%
LiveBench Data Analysis79.2%78.0%
LiveBench IF67.7%60.1%
LiveBench Language82.1%72.6%
LiveBench Reasoning85.8%85.6%
GPQA Diamond (Vals AI)92.4%91.7%
MMLU Pro87.0%86.0%
SimpleQA Verified52.9%41.0%
Maths
FrontierMath (tiers 1-3)64.6%82.1%
AIME-style maths (OTIS mock)98.6%98.3%
LiveBench Mathematics95.1%87.2%
ProofBench v1.150.0%60.0%
Chess puzzles46.0%43.0%
FrontierMath (tier 4)26.8%61.0%
Coding
Terminal-Bench 4.014.1%11.6%
LiveBench Agentic Coding54.9%48.4%
LiveBench Coding77.2%82.9%
Code Migration41.5%44.5%
CyberBench v1.172.9%73.6%
IOI51.6%61.8%
ProgramBench0.0%0.0%
SWE-bench96.4%93.0%
Vibe Code Bench 1-10017.5%22.6%
Vibe Code Bench v1.182.3%77.1%
MirrorCode33.3%21.1%
WeirdML66.2%60.9%
Agents
APEX-Agents47.3%43.0%
SkillsBench53.8%60.5%
SRE Bench1.9%1.5%
Terminal-Bench-Science4.3%0.0%
DTBench93.9%89.1%
LMCA45.5%48.5%
Mystery game puzzles43.0%21.0%
Professional work
Finance Agent (v2)50.4%55.0%
CorpFin v265.4%64.2%
EMB52.8%67.1%
LegalBench82.4%84.0%
Legal Research Bench40.9%36.5%
MedCode42.5%42.4%
MedScribe80.2%84.4%
Tax Agent Bench26.8%24.8%
TaxEval v273.1%76.2%
Novel problems
ARC-AGI-261.3%59.6%
ARC-AGI-190.5%90.7%

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

This page as a chartDownload PNG