Model comparison
DeepSeek-V3 vs Gemini 1.5 Flash (May 2024)
DeepSeek-V3 is the stronger model overall, scoring 39.2 to 32.9 on the Noometry Index.
Last verified . 36 shared benchmarks.
DeepSeek
GoogleSummary
- They share 36 benchmarks with published results for both. DeepSeek-V3 scores higher in 6 categories and Gemini 1.5 Flash (May 2024) in 2 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-V3 leads 37.5 to 26.2.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 37.8% for DeepSeek-V3 and 16.3% for Gemini 1.5 Flash (May 2024).
- DeepSeek-V3 has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3 | Gemini 1.5 Flash (May 2024) | |
|---|---|---|
| Provider | DeepSeek | |
| Noometry Index | 39.2 | 32.9 |
| Released | 2024-12-26 | 2024-05-14 |
| Weights | Open | Proprietary |
| Context window | 164K | — |
| Max output | 164K | — |
| Input $ / M tokens | $0.24 | — |
| Output $ / M tokens | $0.90 | — |
| Results tracked | 60 | 42 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
DeepSeek-V3: 41.9 (#99), Gemini 1.5 Flash (May 2024): 33.9 (#233)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Flash (May 2024) |
|---|---|---|
| WeirdML | 36.1% | 24.9% |
| BigCodeBench Instruct | 50% | 43.5% |
| LMArena Coding | 1368 | 1262 |
| BigCodeBench Complete | 62.2% | 55.1% |
| HumanEval+ | 86.6% | 75.6% |
| MBPP+ | 73% | 67.5% |
| Aider Polyglot | 55.1% | — |
| SciCode | 35.8% | — |
| LiveBench Coding | 70.9% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Gemini 1.5 Flash (May 2024): 26.3 (#97)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Flash (May 2024) |
|---|---|---|
| BALROG | — | 14.6% |
| METR Time Horizons | 49.6% | — |
Reasoning Too close to call
DeepSeek-V3: 20.0 (#226), Gemini 1.5 Flash (May 2024): 20.8 (#212)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Flash (May 2024) |
|---|---|---|
| LMArena Hard Prompts | 1365 | 1257 |
| DTBench | 64.8% | 53.8% |
| Epoch Capabilities Index | 135.95 | 129.37 |
| ForecastBench | 59.1 | 53.9 |
| PIQA | 84.7% | 87.5% |
| SimpleBench | 27.2% | — |
| Kagi LLM Benchmark | 52.3% | — |
| CritPt | 0% | — |
| LiveBench Reasoning | 65.8% | — |
| LiveBench Data Analysis | 60.9% | — |
| LMCA | 15.5% | — |
| BIG-Bench Hard | 87.5% | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| WinoGrande | 85.2% | — |
Math DeepSeek-V3 leads
DeepSeek-V3: 30.4 (#208), Gemini 1.5 Flash (May 2024): 20.8 (#274)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Flash (May 2024) |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 16.3% |
| Omni-MATH | 40.3% | 30.4% |
| LMArena Math | 1373 | 1269 |
| MATH Level 5 | 75.5% | 61.9% |
| FrontierMath (Feb 2025 set) | 1.7% | 0% |
| LiveBench Math | 73.5% | — |
| GSM8K | — | 82.4% |
Knowledge DeepSeek-V3 leads
DeepSeek-V3: 37.5 (#147), Gemini 1.5 Flash (May 2024): 26.2 (#253)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Flash (May 2024) |
|---|---|---|
| GPQA Diamond | 67.6% | 47.3% |
| MMLU-Pro | 72.3% | 67.8% |
| GPQA (HELM) | 53.8% | 43.7% |
| LMArena Expert | 1352 | 1233 |
| MMLU | 87.2% | 77.9% |
| Confabulations | 26.1% | — |
| Vectara Hallucination Rate | 6.1% | — |
| ARC (AI2) Challenge | 95.3% | — |
| BoolQ | — | 85.8% |
| TriviaQA | 82.9% | — |
Multimodal Not comparable
DeepSeek-V3: —, Gemini 1.5 Flash (May 2024): 35.3 (#76)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Flash (May 2024) |
|---|---|---|
| LMArena Vision | — | 1141 |
| Video-MME | — | 70.3% |
| GeoBench | — | 76% |
Multilingual DeepSeek-V3 leads
DeepSeek-V3: 48.8 (#138), Gemini 1.5 Flash (May 2024): 43.1 (#184)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Flash (May 2024) |
|---|---|---|
| LMArena Non-English | 1358 | 1278 |
| LMArena Chinese | 1390 | 1295 |
| LMArena French | 1386 | 1257 |
| LMArena German | 1374 | 1262 |
| LMArena Japanese | 1333 | 1252 |
| LMArena Korean | 1321 | 1220 |
| LMArena Russian | 1372 | 1288 |
| LMArena Spanish | 1358 | 1242 |
Instruction Following DeepSeek-V3 leads
DeepSeek-V3: 73.0 (#125), Gemini 1.5 Flash (May 2024): 67.0 (#199)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Flash (May 2024) |
|---|---|---|
| IFEval | 83.2% | 83.1% |
| LMArena Instruction Following | 1345 | 1258 |
| LiveBench Instruction Following | 81.5% | — |
Long Context Gemini 1.5 Flash (May 2024) leads
DeepSeek-V3: 34.0 (#248), Gemini 1.5 Flash (May 2024): 39.0 (#182)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Flash (May 2024) |
|---|---|---|
| LMArena Longer Query | 1353 | 1284 |
| Fiction.LiveBench | 50% | — |
Writing & Preference DeepSeek-V3 leads
DeepSeek-V3: 57.6 (#124), Gemini 1.5 Flash (May 2024): 48.9 (#190)
| Benchmark | DeepSeek-V3 | Gemini 1.5 Flash (May 2024) |
|---|---|---|
| LMArena Text | 1375 | 1287 |
| LMArena Creative Writing | 1364 | 1285 |
| WildBench | 83% | 79.2% |
| LMArena Multi-Turn | 1389 | 1253 |
| 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 Flash (May 2024)?
DeepSeek-V3 is the stronger model overall, scoring 39.2 to 32.9 on the Noometry Index.
Is DeepSeek-V3 or Gemini 1.5 Flash (May 2024) better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 41.9 versus 33.9 in the Noometry coding category.
How many benchmarks do DeepSeek-V3 and Gemini 1.5 Flash (May 2024) share?
36 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Gemini 1.5 Flash (May 2024) has 42.