Model comparison
DeepSeek V4 Pro vs Gemini 3 Pro
DeepSeek V4 Pro and Gemini 3 Pro score almost the same on the Noometry Index (54.3 vs 54.8), so choose on price, context window or the category you care about most.
Last verified . 36 shared benchmarks.
DeepSeek
GoogleSummary
- They share 36 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 4 categories and Gemini 3 Pro in 5 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek V4 Pro leads 64.8 to 49.9.
- The biggest single-benchmark swing is ARC-AGI-2: 61.3% for DeepSeek V4 Pro and 31.1% for Gemini 3 Pro.
- DeepSeek V4 Pro has downloadable open weights; the other is API-only.
Side by side
| DeepSeek V4 Pro | Gemini 3 Pro | |
|---|---|---|
| Provider | DeepSeek | |
| Noometry Index | 54.3 | 54.8 |
| Released | 2026-04-24 | 2025-11-18 |
| Weights | Open | Proprietary |
| Context window | 1M | — |
| Max output | 393K | — |
| Input $ / M tokens | $0.66 | — |
| Output $ / M tokens | $1.98 | — |
| Results tracked | 48 | 67 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
DeepSeek V4 Pro: 52.4 (#34), Gemini 3 Pro: 51.6 (#39)
| Benchmark | DeepSeek V4 Pro | Gemini 3 Pro |
|---|---|---|
| SWE-bench Verified | 77.6% | 72.9% |
| LMArena WebDev | 1582 | 1440 |
| WeirdML | 66.2% | 69.9% |
| LMArena Coding | 1470 | 1481 |
| ALE-Bench | 1,403 | 1,177 |
| FrontierCode | 28.6% | — |
| SWE-bench Verified (bash only) | — | 74.2% |
| SWE-bench Multilingual | — | 68.7% |
| SciCode | 51% | — |
| GSO | — | 18.6% |
| AlgoTune | — | 1.83 |
Agentic & Tool Use Gemini 3 Pro leads
DeepSeek V4 Pro: 32.8 (#58), Gemini 3 Pro: 40.6 (#23)
| Benchmark | DeepSeek V4 Pro | Gemini 3 Pro |
|---|---|---|
| Vending-Bench 2 | 3,285 | 5,478 |
| Terminal-Bench | — | 69.4% |
| APEX-Agents | 47.3% | — |
| Berkeley Function Calling Leaderboard | — | 72.5% |
| GDPval | — | 40.3% |
| Remote Labor Index | — | 1.3% |
| τ²-bench Airline | — | 80.5% |
| τ²-bench Banking | — | 18% |
| τ²-bench Retail | — | 75.9% |
| τ²-bench Telecom | — | 91% |
| DeepResearch Bench | — | 46.3% |
| BALROG | — | 58.1% |
| LMArena Search | — | 1207 |
| METR Time Horizons | — | 71% |
Reasoning DeepSeek V4 Pro leads
DeepSeek V4 Pro: 56.5 (#24), Gemini 3 Pro: 52.5 (#31)
| Benchmark | DeepSeek V4 Pro | Gemini 3 Pro |
|---|---|---|
| ARC-AGI-2 | 61.3% | 31.1% |
| Kagi LLM Benchmark | 53.5% | 80.1% |
| NYT Connections (extended) | 91.3% | 94.4% |
| ARC-AGI-1 | 90.5% | 75% |
| CritPt | 18% | 6.9% |
| Chess Puzzles | 47% | 31% |
| LMArena Hard Prompts | 1461 | 1480 |
| Epoch Capabilities Index | 155.31 | 152.92 |
| ForecastBench | 56.1 | 61.2 |
| SimpleBench | — | 76.4% |
| EnigmaEval | — | 18.2% |
| Mystery Game Puzzles | 43% | — |
| DTBench | 93.9% | — |
| LMCA | 45.5% | — |
| Surface Evolver Bench | 40% | — |
Math DeepSeek V4 Pro leads
DeepSeek V4 Pro: 64.8 (#30), Gemini 3 Pro: 49.9 (#59)
| Benchmark | DeepSeek V4 Pro | Gemini 3 Pro |
|---|---|---|
| MathArena Final-Answer Competitions | 76.6% | 67% |
| OTIS Mock AIME 2024-2025 | 98.6% | 91.4% |
| ProofBench | 50% | 20% |
| LMArena Math | 1455 | 1476 |
| FrontierMath (Tiers 1-3) | 64.6% | — |
| FrontierMath Tier 4 | 26.8% | — |
| Omni-MATH | — | 55.5% |
| FrontierMath (Feb 2025 set) | — | 37.6% |
| FrontierMath Tier 4 (v1) | — | 18.8% |
Knowledge Gemini 3 Pro leads
DeepSeek V4 Pro: 59.5 (#31), Gemini 3 Pro: 64.4 (#16)
| Benchmark | DeepSeek V4 Pro | Gemini 3 Pro |
|---|---|---|
| GPQA Diamond | 91.7% | 92.6% |
| Vectara Hallucination Rate | 8.6% | 13.6% |
| LMArena Expert | 1464 | 1475 |
| Humanity's Last Exam | — | 37.5% |
| SimpleQA Verified | 52.9% | — |
| MMLU-Pro | — | 90.3% |
| GPQA (HELM) | — | 80.3% |
Multimodal Not comparable
DeepSeek V4 Pro: —, Gemini 3 Pro: 57.6 (#2)
| Benchmark | DeepSeek V4 Pro | Gemini 3 Pro |
|---|---|---|
| LMArena Vision | — | 1305 |
| GeoBench | — | 84% |
| VPCT | — | 91% |
| LMArena Document | — | 1434 |
Multilingual Gemini 3 Pro leads
DeepSeek V4 Pro: 54.4 (#45), Gemini 3 Pro: 56.9 (#16)
| Benchmark | DeepSeek V4 Pro | Gemini 3 Pro |
|---|---|---|
| LMArena Non-English | 1439 | 1474 |
| LMArena Chinese | 1486 | 1523 |
| LMArena French | 1472 | 1492 |
| LMArena German | 1458 | 1515 |
| LMArena Japanese | 1445 | 1510 |
| LMArena Korean | 1447 | 1448 |
| LMArena Russian | 1453 | 1493 |
| LMArena Spanish | 1458 | 1470 |
Instruction Following Too close to call
DeepSeek V4 Pro: 76.1 (#47), Gemini 3 Pro: 76.3 (#45)
| Benchmark | DeepSeek V4 Pro | Gemini 3 Pro |
|---|---|---|
| LMArena Instruction Following | 1448 | 1458 |
| IFEval | — | 87.7% |
Long Context Too close to call
DeepSeek V4 Pro: 45.0 (#51), Gemini 3 Pro: 44.0 (#79)
| Benchmark | DeepSeek V4 Pro | Gemini 3 Pro |
|---|---|---|
| LMArena Longer Query | 1458 | 1471 |
| CL-bench | — | 15.8% |
| CL-bench Life | 13.5% | — |
Writing & Preference Too close to call
DeepSeek V4 Pro: 65.5 (#46), Gemini 3 Pro: 66.4 (#35)
| Benchmark | DeepSeek V4 Pro | Gemini 3 Pro |
|---|---|---|
| LMArena Text | 1451 | 1479 |
| LMArena Creative Writing | 1446 | 1482 |
| EQ-Bench Creative Writing | 1553 | 1525 |
| LMArena Multi-Turn | 1467 | 1484 |
| WildBench | — | 85.9% |
| EQ-Bench 4 | 1166 | — |
Frequently asked questions
Is DeepSeek V4 Pro better than Gemini 3 Pro?
DeepSeek V4 Pro and Gemini 3 Pro score almost the same on the Noometry Index (54.3 vs 54.8), so choose on price, context window or the category you care about most.
Is DeepSeek V4 Pro or Gemini 3 Pro better for coding?
They score almost the same on coding (52.4 vs 51.6); test both on your own repository before choosing.
How many benchmarks do DeepSeek V4 Pro and Gemini 3 Pro share?
36 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and Gemini 3 Pro has 67.