Model comparison
DeepSeek-V2.5 (Sep 2024) vs Gemini 3 Pro
Gemini 3 Pro is the stronger model overall, scoring 54.8 to 37.6 on the Noometry Index.
Last verified . 17 shared benchmarks.
DeepSeek
GoogleSummary
- They share 17 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 0 categories and Gemini 3 Pro in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Gemini 3 Pro leads 64.4 to 34.8.
- DeepSeek-V2.5 (Sep 2024) has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V2.5 (Sep 2024) | Gemini 3 Pro | |
|---|---|---|
| Provider | DeepSeek | |
| Noometry Index | 37.6 | 54.8 |
| Released | 2024-09-06 | 2025-11-18 |
| Weights | Open | Proprietary |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 22 | 67 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemini 3 Pro leads
DeepSeek-V2.5 (Sep 2024): 31.7 (#281), Gemini 3 Pro: 51.6 (#39)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemini 3 Pro |
|---|---|---|
| LMArena Coding | 1309 | 1481 |
| SWE-bench Verified | — | 72.9% |
| SWE-bench Verified (bash only) | — | 74.2% |
| Aider Polyglot | 17.8% | — |
| LMArena WebDev | — | 1440 |
| SWE-bench Multilingual | — | 68.7% |
| GSO | — | 18.6% |
| WeirdML | — | 69.9% |
| BigCodeBench Instruct | 48.6% | — |
| BigCodeBench Complete | 53.2% | — |
| ALE-Bench | — | 1,177 |
| AlgoTune | — | 1.83 |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |
Agentic & Tool Use Not comparable
DeepSeek-V2.5 (Sep 2024): —, Gemini 3 Pro: 40.6 (#23)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemini 3 Pro |
|---|---|---|
| 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% |
| DeepResearch Bench | — | 46.3% |
| BALROG | — | 58.1% |
| LMArena Search | — | 1207 |
| METR Time Horizons | — | 71% |
| Vending-Bench 2 | — | 5,478 |
Reasoning Gemini 3 Pro leads
DeepSeek-V2.5 (Sep 2024): 25.6 (#145), Gemini 3 Pro: 52.5 (#31)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemini 3 Pro |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1480 |
| ARC-AGI-2 | — | 31.1% |
| SimpleBench | — | 76.4% |
| Kagi LLM Benchmark | — | 80.1% |
| NYT Connections (extended) | — | 94.4% |
| ARC-AGI-1 | — | 75% |
| CritPt | — | 6.9% |
| Chess Puzzles | — | 31% |
| EnigmaEval | — | 18.2% |
| Epoch Capabilities Index | — | 152.92 |
| ForecastBench | — | 61.2 |
Math Gemini 3 Pro leads
DeepSeek-V2.5 (Sep 2024): 35.9 (#177), Gemini 3 Pro: 49.9 (#59)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemini 3 Pro |
|---|---|---|
| LMArena Math | 1288 | 1476 |
| MathArena Final-Answer Competitions | — | 67% |
| OTIS Mock AIME 2024-2025 | — | 91.4% |
| ProofBench | — | 20% |
| Omni-MATH | — | 55.5% |
| FrontierMath (Feb 2025 set) | — | 37.6% |
| FrontierMath Tier 4 (v1) | — | 18.8% |
Knowledge Gemini 3 Pro leads
DeepSeek-V2.5 (Sep 2024): 34.8 (#193), Gemini 3 Pro: 64.4 (#16)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemini 3 Pro |
|---|---|---|
| LMArena Expert | 1266 | 1475 |
| GPQA Diamond | — | 92.6% |
| Humanity's Last Exam | — | 37.5% |
| MMLU-Pro | — | 90.3% |
| Vectara Hallucination Rate | — | 13.6% |
| GPQA (HELM) | — | 80.3% |
Multimodal Not comparable
DeepSeek-V2.5 (Sep 2024): —, Gemini 3 Pro: 57.6 (#2)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemini 3 Pro |
|---|---|---|
| LMArena Vision | — | 1305 |
| GeoBench | — | 84% |
| VPCT | — | 91% |
| LMArena Document | — | 1434 |
Multilingual Gemini 3 Pro leads
DeepSeek-V2.5 (Sep 2024): 42.5 (#193), Gemini 3 Pro: 56.9 (#16)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemini 3 Pro |
|---|---|---|
| LMArena Non-English | 1273 | 1474 |
| LMArena Chinese | 1318 | 1523 |
| LMArena French | 1289 | 1492 |
| LMArena German | 1258 | 1515 |
| LMArena Japanese | 1228 | 1510 |
| LMArena Korean | 1209 | 1448 |
| LMArena Russian | 1289 | 1493 |
| LMArena Spanish | 1248 | 1470 |
Instruction Following Gemini 3 Pro leads
DeepSeek-V2.5 (Sep 2024): 67.5 (#194), Gemini 3 Pro: 76.3 (#45)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemini 3 Pro |
|---|---|---|
| LMArena Instruction Following | 1280 | 1458 |
| IFEval | — | 87.7% |
Long Context Gemini 3 Pro leads
DeepSeek-V2.5 (Sep 2024): 39.5 (#174), Gemini 3 Pro: 44.0 (#79)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemini 3 Pro |
|---|---|---|
| LMArena Longer Query | 1301 | 1471 |
| CL-bench | — | 15.8% |
Writing & Preference Gemini 3 Pro leads
DeepSeek-V2.5 (Sep 2024): 49.8 (#187), Gemini 3 Pro: 66.4 (#35)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemini 3 Pro |
|---|---|---|
| LMArena Text | 1294 | 1479 |
| LMArena Creative Writing | 1285 | 1482 |
| LMArena Multi-Turn | 1297 | 1484 |
| EQ-Bench Creative Writing | — | 1525 |
| WildBench | — | 85.9% |
Frequently asked questions
Is DeepSeek-V2.5 (Sep 2024) better than Gemini 3 Pro?
Gemini 3 Pro is the stronger model overall, scoring 54.8 to 37.6 on the Noometry Index.
Is DeepSeek-V2.5 (Sep 2024) or Gemini 3 Pro better for coding?
Gemini 3 Pro scores higher on coding benchmarks: 51.6 versus 31.7 in the Noometry coding category.
How many benchmarks do DeepSeek-V2.5 (Sep 2024) and Gemini 3 Pro share?
17 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and Gemini 3 Pro has 67.