Model comparison
DeepSeek-R1 vs Gemini 3 Pro
Gemini 3 Pro is the stronger model overall, scoring 54.8 to 42.3 on the Noometry Index.
Last verified . 39 shared benchmarks.
DeepSeek
GoogleSummary
- They share 39 benchmarks with published results for both. DeepSeek-R1 scores higher in 1 category and Gemini 3 Pro in 8 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Gemini 3 Pro leads 52.5 to 18.6.
- The biggest single-benchmark swing is ARC-AGI-1: 21.2% for DeepSeek-R1 and 75% for Gemini 3 Pro.
Side by side
| DeepSeek-R1 | Gemini 3 Pro | |
|---|---|---|
| Provider | DeepSeek | |
| Noometry Index | 42.3 | 54.8 |
| Released | 2025-01-20 | 2025-11-18 |
| Weights | Proprietary | Proprietary |
| Context window | 164K | — |
| Max output | 64K | — |
| Input $ / M tokens | $0.50 | — |
| Output $ / M tokens | $2.15 | — |
| Results tracked | 52 | 67 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemini 3 Pro leads
DeepSeek-R1: 46.3 (#68), Gemini 3 Pro: 51.6 (#39)
| Benchmark | DeepSeek-R1 | Gemini 3 Pro |
|---|---|---|
| WeirdML | 41.6% | 69.9% |
| LMArena Coding | 1427 | 1481 |
| ALE-Bench | 804.12 | 1,177 |
| AlgoTune | 1.7 | 1.83 |
| SWE-bench Verified | — | 72.9% |
| SWE-bench Verified (bash only) | — | 74.2% |
| Aider Polyglot | 71.4% | — |
| LMArena WebDev | — | 1440 |
| SWE-bench Multilingual | — | 68.7% |
| SciCode | 35.7% | — |
| GSO | — | 18.6% |
| LiveBench Coding | 66.7% | — |
Agentic & Tool Use Gemini 3 Pro leads
DeepSeek-R1: 30.7 (#75), Gemini 3 Pro: 40.6 (#23)
| Benchmark | DeepSeek-R1 | Gemini 3 Pro |
|---|---|---|
| DeepResearch Bench | 35.1% | 46.3% |
| BALROG | 34.9% | 58.1% |
| METR Time Horizons | 53.8% | 71% |
| Terminal-Bench | — | 69.4% |
| 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% |
| LMArena Search | — | 1207 |
| Vending-Bench 2 | — | 5,478 |
Reasoning Gemini 3 Pro leads
DeepSeek-R1: 18.6 (#278), Gemini 3 Pro: 52.5 (#31)
| Benchmark | DeepSeek-R1 | Gemini 3 Pro |
|---|---|---|
| ARC-AGI-2 | 1.3% | 31.1% |
| SimpleBench | 40.8% | 76.4% |
| Kagi LLM Benchmark | 69.4% | 80.1% |
| ARC-AGI-1 | 21.2% | 75% |
| CritPt | 1.1% | 6.9% |
| LMArena Hard Prompts | 1416 | 1480 |
| Epoch Capabilities Index | 141.29 | 152.92 |
| ForecastBench | 60 | 61.2 |
| NYT Connections (extended) | — | 94.4% |
| Chess Puzzles | — | 31% |
| EnigmaEval | — | 18.2% |
| LiveBench Reasoning | 83.2% | — |
| LiveBench Data Analysis | 69.8% | — |
| LiveBench | 71.6% | — |
Math Gemini 3 Pro leads
DeepSeek-R1: 43.8 (#79), Gemini 3 Pro: 49.9 (#59)
| Benchmark | DeepSeek-R1 | Gemini 3 Pro |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 91.4% |
| Omni-MATH | 42.4% | 55.5% |
| LMArena Math | 1400 | 1476 |
| MathArena Final-Answer Competitions | — | 67% |
| ProofBench | — | 20% |
| LiveBench Math | 80.7% | — |
| MATH Level 5 | 96.6% | — |
| FrontierMath (Feb 2025 set) | — | 37.6% |
| FrontierMath Tier 4 (v1) | — | 18.8% |
Knowledge Gemini 3 Pro leads
DeepSeek-R1: 44.5 (#87), Gemini 3 Pro: 64.4 (#16)
| Benchmark | DeepSeek-R1 | Gemini 3 Pro |
|---|---|---|
| GPQA Diamond | 76.3% | 92.6% |
| MMLU-Pro | 79.3% | 90.3% |
| Vectara Hallucination Rate | 11.3% | 13.6% |
| GPQA (HELM) | 66.6% | 80.3% |
| LMArena Expert | 1394 | 1475 |
| Humanity's Last Exam | — | 37.5% |
| Confabulations | 12.7% | — |
Multimodal Not comparable
DeepSeek-R1: —, Gemini 3 Pro: 57.6 (#2)
| Benchmark | DeepSeek-R1 | Gemini 3 Pro |
|---|---|---|
| LMArena Vision | — | 1305 |
| GeoBench | — | 84% |
| VPCT | — | 91% |
| LMArena Document | — | 1434 |
Multilingual Gemini 3 Pro leads
DeepSeek-R1: 52.4 (#85), Gemini 3 Pro: 56.9 (#16)
| Benchmark | DeepSeek-R1 | Gemini 3 Pro |
|---|---|---|
| LMArena Non-English | 1412 | 1474 |
| LMArena Chinese | 1442 | 1523 |
| LMArena French | 1417 | 1492 |
| LMArena German | 1404 | 1515 |
| LMArena Japanese | 1391 | 1510 |
| LMArena Korean | 1360 | 1448 |
| LMArena Russian | 1423 | 1493 |
| LMArena Spanish | 1411 | 1470 |
Instruction Following Gemini 3 Pro leads
DeepSeek-R1: 72.0 (#143), Gemini 3 Pro: 76.3 (#45)
| Benchmark | DeepSeek-R1 | Gemini 3 Pro |
|---|---|---|
| IFEval | 78.4% | 87.7% |
| LMArena Instruction Following | 1382 | 1458 |
| LiveBench Instruction Following | 80.5% | — |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), Gemini 3 Pro: 44.0 (#79)
| Benchmark | DeepSeek-R1 | Gemini 3 Pro |
|---|---|---|
| LMArena Longer Query | 1391 | 1471 |
| Fiction.LiveBench | 75% | — |
| CL-bench | — | 15.8% |
Writing & Preference Gemini 3 Pro leads
DeepSeek-R1: 61.4 (#88), Gemini 3 Pro: 66.4 (#35)
| Benchmark | DeepSeek-R1 | Gemini 3 Pro |
|---|---|---|
| LMArena Text | 1428 | 1479 |
| LMArena Creative Writing | 1405 | 1482 |
| EQ-Bench Creative Writing | 1500 | 1525 |
| WildBench | 82.8% | 85.9% |
| LMArena Multi-Turn | 1405 | 1484 |
| Short-Story Creative Writing | 83% | — |
| LiveBench Language | 48.5% | — |
Frequently asked questions
Is DeepSeek-R1 better than Gemini 3 Pro?
Gemini 3 Pro is the stronger model overall, scoring 54.8 to 42.3 on the Noometry Index.
Is DeepSeek-R1 or Gemini 3 Pro better for coding?
Gemini 3 Pro scores higher on coding benchmarks: 51.6 versus 46.3 in the Noometry coding category.
How many benchmarks do DeepSeek-R1 and Gemini 3 Pro share?
39 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and Gemini 3 Pro has 67.