Compare · ModelsLive · 2 picked · head to head
DeepSeek V3 vs Gemini 3 Pro
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Gemini 3 Pro wins on 11/11 benchmarks
Gemini 3 Pro wins 11 of 11 shared benchmarks. Leads in arena · math · knowledge.
Category leads
arena·Gemini 3 Promath·Gemini 3 Proknowledge·Gemini 3 Prolanguage·Gemini 3 Proreasoning·Gemini 3 Procoding·Gemini 3 Pro
Hype vs Reality
Attention vs performance
DeepSeek V3
#43 by perf·no signal
Gemini 3 Pro
#38 by perf·no signal
Vendor risk
Mixed exposure
One or more vendors flagged
DeepSeek
$3.4B·Tier 1
Google DeepMind
$4.00T·Tier 1
Head to head
11 benchmarks · 2 models
DeepSeek V3Gemini 3 Pro
Chatbot Arena Elo · Overall
Gemini 3 Pro leads by +128.0
DeepSeek V3
1358.2
Gemini 3 Pro
1486.2
FrontierMath-2025-02-28-Private
Gemini 3 Pro leads by +35.9
FrontierMath (Feb 2025) · original research-level math problems created by mathematicians, testing capabilities at the boundary of current AI mathematical reasoning.
DeepSeek V3
1.7
Gemini 3 Pro
37.6
GPQA diamond
Gemini 3 Pro leads by +48.1
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
DeepSeek V3
42.0
Gemini 3 Pro
90.2
HELM · GPQA
Gemini 3 Pro leads by +26.5
DeepSeek V3
53.8
Gemini 3 Pro
80.3
HELM · IFEval
Gemini 3 Pro leads by +4.4
DeepSeek V3
83.2
Gemini 3 Pro
87.6
HELM · MMLU-Pro
Gemini 3 Pro leads by +18.0
DeepSeek V3
72.3
Gemini 3 Pro
90.3
HELM · Omni-MATH
Gemini 3 Pro leads by +15.3
DeepSeek V3
40.3
Gemini 3 Pro
55.6
HELM · WildBench
Gemini 3 Pro leads by +2.8
DeepSeek V3
83.1
Gemini 3 Pro
85.9
OTIS Mock AIME 2024-2025
Gemini 3 Pro leads by +75.6
OTIS Mock AIME 2024–2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
DeepSeek V3
15.8
Gemini 3 Pro
91.4
SimpleBench
Gemini 3 Pro leads by +69.0
SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking.
DeepSeek V3
2.7
Gemini 3 Pro
71.7
WeirdML
Gemini 3 Pro leads by +33.9
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
DeepSeek V3
36.1
Gemini 3 Pro
69.9
Full benchmark table
| Benchmark | DeepSeek V3 | Gemini 3 Pro |
|---|---|---|
Chatbot Arena Elo · Overall | 1358.2 | 1486.2 |
FrontierMath-2025-02-28-Private FrontierMath (Feb 2025) · original research-level math problems created by mathematicians, testing capabilities at the boundary of current AI mathematical reasoning. | 1.7 | 37.6 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 42.0 | 90.2 |
HELM · GPQA | 53.8 | 80.3 |
HELM · IFEval | 83.2 | 87.6 |
HELM · MMLU-Pro | 72.3 | 90.3 |
HELM · Omni-MATH | 40.3 | 55.6 |
HELM · WildBench | 83.1 | 85.9 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024–2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 15.8 | 91.4 |
SimpleBench SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking. | 2.7 | 71.7 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 36.1 | 69.9 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| DeepSeek V3 | $0.32 | $0.89 | 164K tokens (~82 books) | $4.63 |
| Gemini 3 Pro | — | — | — | — |