Model comparison
DeepSeek-V3 vs Gemini 1.5 Pro (May 2024)
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 32.1 on the Noometry Index.
Last verified . 37 shared benchmarks.
DeepSeek
GoogleSummary
- They share 37 benchmarks with published results for both. DeepSeek-V3 scores higher in 7 categories and Gemini 1.5 Pro (May 2024) in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek-V3 leads 20.5 to 12.3.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 37.8% for DeepSeek-V3 and 23.1% for Gemini 1.5 Pro (May 2024).
- DeepSeek-V3 has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3 | Gemini 1.5 Pro (May 2024) | |
|---|---|---|
| Provider | DeepSeek | |
| Noometry Index | 39.5 | 32.1 |
| Released | 2024-12-26 | 2024-02-15 |
| Weights | Open | Proprietary |
| Context window | 164K | — |
| Max output | 164K | — |
| Input $ / M tokens | $0.24 | — |
| Output $ / M tokens | $0.90 | — |
| Results tracked | 60 | 45 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
DeepSeek-V3: 42.3 (#106), Gemini 1.5 Pro (May 2024): 34.2 (#241)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Pro (May 2024) |
|---|---|---|
| WeirdML | 36.1% | 22.2% |
| BigCodeBench Instruct | 50% | 43.8% |
| LMArena Coding | 1368 | 1294 |
| BigCodeBench Complete | 62.2% | 57.5% |
| HumanEval+ | 86.6% | 79.3% |
| MBPP+ | 73% | 74.6% |
| Aider Polyglot | 55.1% | — |
| SciCode | 35.8% | — |
| LiveBench Coding | 70.9% | — |
| CadEval | — | 34% |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Gemini 1.5 Pro (May 2024): 17.9 (#145)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Pro (May 2024) |
|---|---|---|
| TheAgentCompany | — | 3.4% |
| Cybench | — | 7.5% |
| BALROG | — | 21% |
| METR Time Horizons | 49.6% | — |
Reasoning DeepSeek-V3 leads
DeepSeek-V3: 20.5 (#236), Gemini 1.5 Pro (May 2024): 12.3 (#338)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Pro (May 2024) |
|---|---|---|
| SimpleBench | 27.2% | 27.1% |
| LMArena Hard Prompts | 1365 | 1296 |
| DTBench | 64.8% | 59% |
| BIG-Bench Hard | 87.5% | 89.2% |
| Epoch Capabilities Index | 135.94 | 131.73 |
| ForecastBench | 59.1 | 58.4 |
| ARC-AGI-2 | — | 0.8% |
| Kagi LLM Benchmark | 52.3% | — |
| CritPt | 0% | — |
| LiveBench Reasoning | 65.8% | — |
| LiveBench Data Analysis | 60.9% | — |
| LMCA | 15.5% | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math DeepSeek-V3 leads
DeepSeek-V3: 32.1 (#219), Gemini 1.5 Pro (May 2024): 25.8 (#266)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Pro (May 2024) |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 23.1% |
| Omni-MATH | 40.3% | 36.4% |
| LMArena Math | 1373 | 1315 |
| MATH Level 5 | 75.5% | 70.4% |
| LiveBench Math | 73.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |
Knowledge DeepSeek-V3 leads
DeepSeek-V3: 37.5 (#155), Gemini 1.5 Pro (May 2024): 29.4 (#239)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Pro (May 2024) |
|---|---|---|
| GPQA Diamond | 67.6% | 57.2% |
| MMLU-Pro | 72.3% | 73.7% |
| Confabulations | 26.1% | 13.5% |
| GPQA (HELM) | 53.8% | 53.4% |
| LMArena Expert | 1351 | 1279 |
| MMLU | 87.2% | 86.9% |
| Humanity's Last Exam | — | 4.6% |
| Vectara Hallucination Rate | 6.1% | — |
| ARC (AI2) Challenge | 95.3% | — |
| TriviaQA | 82.9% | — |
Multimodal Not comparable
DeepSeek-V3: —, Gemini 1.5 Pro (May 2024): 36.8 (#77)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Pro (May 2024) |
|---|---|---|
| LMArena Vision | — | 1161 |
| Video-MME | — | 75% |
Multilingual DeepSeek-V3 leads
DeepSeek-V3: 48.5 (#143), Gemini 1.5 Pro (May 2024): 45.3 (#174)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Pro (May 2024) |
|---|---|---|
| LMArena Non-English | 1358 | 1312 |
| LMArena Chinese | 1391 | 1331 |
| LMArena French | 1385 | 1302 |
| LMArena German | 1374 | 1286 |
| LMArena Japanese | 1333 | 1292 |
| LMArena Korean | 1319 | 1298 |
| LMArena Russian | 1373 | 1320 |
| LMArena Spanish | 1358 | 1311 |
Instruction Following DeepSeek-V3 leads
DeepSeek-V3: 72.8 (#130), Gemini 1.5 Pro (May 2024): 68.6 (#185)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Pro (May 2024) |
|---|---|---|
| IFEval | 83.2% | 83.7% |
| LMArena Instruction Following | 1345 | 1297 |
| LiveBench Instruction Following | 81.5% | — |
Long Context Gemini 1.5 Pro (May 2024) leads
DeepSeek-V3: 34.0 (#253), Gemini 1.5 Pro (May 2024): 39.8 (#169)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Pro (May 2024) |
|---|---|---|
| LMArena Longer Query | 1352 | 1308 |
| Fiction.LiveBench | 50% | — |
Writing & Preference DeepSeek-V3 leads
DeepSeek-V3: 57.4 (#130), Gemini 1.5 Pro (May 2024): 52.4 (#172)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Pro (May 2024) |
|---|---|---|
| LMArena Text | 1375 | 1319 |
| LMArena Creative Writing | 1364 | 1333 |
| WildBench | 83% | 81.3% |
| LMArena Multi-Turn | 1389 | 1296 |
| Short-Story Creative Writing | 77% | — |
| EQ-Bench Creative Writing | 1472 | — |
| LiveBench Language | 49.1% | — |
Frequently asked questions
Is DeepSeek-V3 better than Gemini 1.5 Pro (May 2024)?
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 32.1 on the Noometry Index.
Is DeepSeek-V3 or Gemini 1.5 Pro (May 2024) better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 34.2 in the Noometry coding category.
How many benchmarks do DeepSeek-V3 and Gemini 1.5 Pro (May 2024) share?
37 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Gemini 1.5 Pro (May 2024) has 45.