Model comparison
DeepSeek-V3 vs Gemini 2.0 Flash-Lite
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 37.8 on the Noometry Index.
Last verified . 31 shared benchmarks.
DeepSeek
GoogleSummary
- They share 31 benchmarks with published results for both. DeepSeek-V3 scores higher in 5 categories and Gemini 2.0 Flash-Lite in 3 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in long context, where Gemini 2.0 Flash-Lite leads 40.1 to 34.0.
- The biggest single-benchmark swing is LiveBench Coding: 70.9% for DeepSeek-V3 and 47.1% for Gemini 2.0 Flash-Lite.
- DeepSeek-V3 has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3 | Gemini 2.0 Flash-Lite | |
|---|---|---|
| Provider | DeepSeek | |
| Noometry Index | 39.5 | 37.8 |
| Released | 2024-12-26 | 2025-02-05 |
| Weights | Open | Proprietary |
| Context window | 164K | — |
| Max output | 164K | — |
| Input $ / M tokens | $0.24 | — |
| Output $ / M tokens | $0.90 | — |
| Results tracked | 60 | 32 |
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 2.0 Flash-Lite: 37.9 (#185)
| Benchmark | DeepSeek-V3 | Gemini 2.0 Flash-Lite |
|---|---|---|
| LiveBench Coding | 70.9% | 47.1% |
| LMArena Coding | 1368 | 1322 |
| Aider Polyglot | 55.1% | — |
| SciCode | 35.8% | — |
| WeirdML | 36.1% | — |
| BigCodeBench Instruct | 50% | — |
| BigCodeBench Complete | 62.2% | — |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Gemini 2.0 Flash-Lite: —
| Benchmark | DeepSeek-V3 | Gemini 2.0 Flash-Lite |
|---|---|---|
| METR Time Horizons | 49.6% | — |
Reasoning Gemini 2.0 Flash-Lite leads
DeepSeek-V3: 20.5 (#236), Gemini 2.0 Flash-Lite: 22.0 (#210)
| Benchmark | DeepSeek-V3 | Gemini 2.0 Flash-Lite |
|---|---|---|
| LiveBench Reasoning | 65.8% | 50.1% |
| LMArena Hard Prompts | 1365 | 1324 |
| DTBench | 64.8% | 52.5% |
| LiveBench Data Analysis | 60.9% | 65.5% |
| ForecastBench | 59.1 | 57.1 |
| LiveBench | 66.9% | 54.3% |
| SimpleBench | 27.2% | — |
| Kagi LLM Benchmark | 52.3% | — |
| CritPt | 0% | — |
| LMCA | 15.5% | — |
| BIG-Bench Hard | 87.5% | — |
| Epoch Capabilities Index | 135.94 | — |
| HellaSwag | 88.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math Gemini 2.0 Flash-Lite leads
DeepSeek-V3: 32.1 (#219), Gemini 2.0 Flash-Lite: 34.1 (#196)
| Benchmark | DeepSeek-V3 | Gemini 2.0 Flash-Lite |
|---|---|---|
| Omni-MATH | 40.3% | 37.4% |
| LiveBench Math | 73.5% | 58.1% |
| LMArena Math | 1373 | 1309 |
| OTIS Mock AIME 2024-2025 | 37.8% | — |
| MATH Level 5 | 75.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |
Knowledge DeepSeek-V3 leads
DeepSeek-V3: 37.5 (#155), Gemini 2.0 Flash-Lite: 35.0 (#189)
| Benchmark | DeepSeek-V3 | Gemini 2.0 Flash-Lite |
|---|---|---|
| MMLU-Pro | 72.3% | 72% |
| GPQA (HELM) | 53.8% | 50% |
| LMArena Expert | 1351 | 1305 |
| GPQA Diamond | 67.6% | — |
| Confabulations | 26.1% | — |
| Vectara Hallucination Rate | 6.1% | — |
| ARC (AI2) Challenge | 95.3% | — |
| MMLU | 87.2% | — |
| TriviaQA | 82.9% | — |
Multimodal Not comparable
DeepSeek-V3: —, Gemini 2.0 Flash-Lite: 31.2 (#109)
| Benchmark | DeepSeek-V3 | Gemini 2.0 Flash-Lite |
|---|---|---|
| LMArena Vision | — | 1100 |
Multilingual DeepSeek-V3 leads
DeepSeek-V3: 48.5 (#143), Gemini 2.0 Flash-Lite: 46.0 (#161)
| Benchmark | DeepSeek-V3 | Gemini 2.0 Flash-Lite |
|---|---|---|
| LMArena Non-English | 1358 | 1323 |
| LMArena Chinese | 1391 | 1339 |
| LMArena French | 1385 | 1347 |
| LMArena German | 1374 | 1306 |
| LMArena Japanese | 1333 | 1301 |
| LMArena Korean | 1319 | 1325 |
| LMArena Russian | 1373 | 1328 |
| LMArena Spanish | 1358 | 1313 |
Instruction Following DeepSeek-V3 leads
DeepSeek-V3: 72.8 (#130), Gemini 2.0 Flash-Lite: 70.4 (#163)
| Benchmark | DeepSeek-V3 | Gemini 2.0 Flash-Lite |
|---|---|---|
| LiveBench Instruction Following | 81.5% | 78.3% |
| IFEval | 83.2% | 82.4% |
| LMArena Instruction Following | 1345 | 1305 |
Long Context Gemini 2.0 Flash-Lite leads
DeepSeek-V3: 34.0 (#253), Gemini 2.0 Flash-Lite: 40.1 (#160)
| Benchmark | DeepSeek-V3 | Gemini 2.0 Flash-Lite |
|---|---|---|
| LMArena Longer Query | 1352 | 1320 |
| Fiction.LiveBench | 50% | — |
Writing & Preference DeepSeek-V3 leads
DeepSeek-V3: 57.4 (#130), Gemini 2.0 Flash-Lite: 51.7 (#177)
| Benchmark | DeepSeek-V3 | Gemini 2.0 Flash-Lite |
|---|---|---|
| LMArena Text | 1375 | 1330 |
| LMArena Creative Writing | 1364 | 1319 |
| WildBench | 83% | 79% |
| LMArena Multi-Turn | 1389 | 1307 |
| LiveBench Language | 49.1% | 34.3% |
| Short-Story Creative Writing | 77% | — |
| EQ-Bench Creative Writing | 1472 | — |
Frequently asked questions
Is DeepSeek-V3 better than Gemini 2.0 Flash-Lite?
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 37.8 on the Noometry Index.
Is DeepSeek-V3 or Gemini 2.0 Flash-Lite better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 37.9 in the Noometry coding category.
How many benchmarks do DeepSeek-V3 and Gemini 2.0 Flash-Lite share?
31 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Gemini 2.0 Flash-Lite has 32.