Skip to content
VECTOR WIREAI INTELLIGENCE
UTC

DeepSeek-V4-Pro vs GPT-5.6 Luna

43 SHARED BENCHMARKS

Across 43 shared benchmarks, DeepSeek-V4-Pro scores higher on 10 and GPT-5.6 Luna on 33. The widest gap is DeepSWE, where GPT-5.6 Luna scores 67.2 against 12.8. Tracked API pricing per million tokens: DeepSeek-V4-Pro $0.43 in / $0.87 out, GPT-5.6 Luna $0.20 in / $1.20 out.

DEEPSEEKVSOPENAI43 SHARED10–33 HEAD-TO-HEADNEWEST SCORE ADDED

At a glance

MakerDeepSeekOpenAI
Released
Price per 1M tokens input / output$0.43 / $0.87$0.20 / $1.20
Cost of 1M in + 1M out$1.30 1.1× less$1.40
Head-to-head of 43 shared benchmarks10 wins33 wins
Scores tracked independently verified256 27 ◆201 43 ◆

Release dates per Artificial Analysis. Prices: Artificial Analysis for DeepSeek-V4-Pro; OpenAI's own price page for GPT-5.6 Luna. ◆ = independently verified score.

Where each leadsby capability area · benchmark wins

AreaDeepSeek-V4-ProWINSGPT-5.6 Luna
Reasoning14GPT-5.6 Luna leads 4 of 5 · widest: CritPt 12.9 vs 20.6
Coding07GPT-5.6 Luna leads 7 of 7 · widest: Terminal-Bench 2.1 64 vs 80.9
Agentic24GPT-5.6 Luna leads 4 of 6 · widest: AA ApexAgents 24.3 vs 35.8
Factuality21DeepSeek-V4-Pro leads 2 of 3 · widest: SimpleQA Verified 46.2 vs 41
Instruction Following10DeepSeek-V4-Pro leads 1 of 1
Long Context01GPT-5.6 Luna leads 1 of 1 · widest: AA-LCR 74.7 vs 83.7
Math14GPT-5.6 Luna leads 4 of 5 · widest: FrontierMath Tiers 1-3 (v2) 45.3 vs 82.1

Biggest gaps

DeepSeek-V4-Pro pulls furthest ahead on

  1. Terminal-Bench 4.014.6 vs 11.6
  2. SimpleQA Verified46.2 vs 41
  3. SimpleBench50.9 vs 46.8

GPT-5.6 Luna pulls furthest ahead on

  1. FrontierMath Tiers 1-3 (v2)82.1 vs 45.3
  2. CritPt20.6 vs 12.9
  3. AA ApexAgents35.8 vs 24.3

Every shared benchmark43 · grouped by area

Reasoning 5

CritPt12.920.6
GPQA Diamond88.891.1
LiveBench · Reasoning82.7 ◆◆ 85.6
SimpleBench50.9 ◆◆ 46.8

Agentic 6

BrowseComp83.483.3
GDPVal32.247.2
Terminal-Bench 4.014.611.6

Instruction Following 1

Long Context 1

AA-LCR74.783.7

Math 5

Other shared benchmarks 15

Each score is the one the model's own page shows — the most authoritative tracked result for that benchmark, on the benchmark's standard methodology (source-attributed). ↓ marks lower-is-better metrics; ◆ an independently verified score. Open either model for its full surface, provenance and pricing. Quoted rates are the price-setter row we currently track for each model — its direct or vendor-official listing where one exists (direct, openai-official), otherwise the lowest tracked offer.

Compare DeepSeek-V4-Pro withALL PAIRINGS →

Compare GPT-5.6 Luna withALL PAIRINGS →