Model comparison
DeepSeek-V3 vs Gemini 4 Argon
Gemini 4 Argon is the stronger model overall, scoring 56.5 to 39.5 on the Noometry Index.
Last verified . 15 shared benchmarks.
DeepSeek
GoogleSummary
- They share 15 benchmarks with published results for both. DeepSeek-V3 scores higher in 0 categories and Gemini 4 Argon in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Gemini 4 Argon leads 73.9 to 32.1.
- The biggest single-benchmark swing is CritPt: 0% for DeepSeek-V3 and 27.1% for Gemini 4 Argon.
- DeepSeek-V3 has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3 | Gemini 4 Argon | |
|---|---|---|
| Provider | DeepSeek | |
| Noometry Index | 39.5 | 56.5 |
| Released | 2024-12-26 | 2026-09-30 |
| Weights | Open | Proprietary |
| Context window | 164K | — |
| Max output | 164K | — |
| Input $ / M tokens | $0.24 | — |
| Output $ / M tokens | $0.90 | — |
| Results tracked | 60 | 21 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemini 4 Argon leads
DeepSeek-V3: 42.3 (#106), Gemini 4 Argon: 62.0 (#9)
| Benchmark | DeepSeek-V3 | Gemini 4 Argon |
|---|---|---|
| SciCode | 35.8% | 61.8% |
| LMArena Coding | 1368 | 1550 |
| Aider Polyglot | 55.1% | — |
| LMArena WebDev | — | 1678 |
| FrontierSWE | — | 55% |
| WeirdML | 36.1% | — |
| BigCodeBench Instruct | 50% | — |
| LiveBench Coding | 70.9% | — |
| BigCodeBench Complete | 62.2% | — |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Gemini 4 Argon: 49.2 (#8)
| Benchmark | DeepSeek-V3 | Gemini 4 Argon |
|---|---|---|
| APEX-Agents | — | 82.2% |
| METR Time Horizons | 49.6% | — |
| Vending-Bench 2 | — | 13,718 |
Reasoning Gemini 4 Argon leads
DeepSeek-V3: 20.5 (#236), Gemini 4 Argon: 44.2 (#47)
| Benchmark | DeepSeek-V3 | Gemini 4 Argon |
|---|---|---|
| CritPt | 0% | 27.1% |
| LMArena Hard Prompts | 1365 | 1548 |
| SimpleBench | 27.2% | — |
| Kagi LLM Benchmark | 52.3% | — |
| LiveBench Reasoning | 65.8% | — |
| DTBench | 64.8% | — |
| LiveBench Data Analysis | 60.9% | — |
| LMCA | 15.5% | — |
| BIG-Bench Hard | 87.5% | — |
| Epoch Capabilities Index | 135.94 | — |
| ForecastBench | 59.1 | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math Gemini 4 Argon leads
DeepSeek-V3: 32.1 (#219), Gemini 4 Argon: 73.9 (#17)
| Benchmark | DeepSeek-V3 | Gemini 4 Argon |
|---|---|---|
| LMArena Math | 1373 | 1536 |
| OTIS Mock AIME 2024-2025 | 37.8% | — |
| ProofBench | — | 99% |
| Omni-MATH | 40.3% | — |
| LiveBench Math | 73.5% | — |
| MATH Level 5 | 75.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |
Knowledge Gemini 4 Argon leads
DeepSeek-V3: 37.5 (#155), Gemini 4 Argon: 43.9 (#89)
| Benchmark | DeepSeek-V3 | Gemini 4 Argon |
|---|---|---|
| LMArena Expert | 1351 | 1553 |
| GPQA Diamond | 67.6% | — |
| MMLU-Pro | 72.3% | — |
| Confabulations | 26.1% | — |
| Vectara Hallucination Rate | 6.1% | — |
| GPQA (HELM) | 53.8% | — |
| ARC (AI2) Challenge | 95.3% | — |
| MMLU | 87.2% | — |
| TriviaQA | 82.9% | — |
Multimodal Not comparable
DeepSeek-V3: —, Gemini 4 Argon: 46.5 (#14)
| Benchmark | DeepSeek-V3 | Gemini 4 Argon |
|---|---|---|
| Blueprint-Bench 2 | — | 54.4% |
Multilingual Gemini 4 Argon leads
DeepSeek-V3: 48.5 (#143), Gemini 4 Argon: 60.3 (#1)
| Benchmark | DeepSeek-V3 | Gemini 4 Argon |
|---|---|---|
| LMArena Non-English | 1358 | 1523 |
| LMArena Chinese | 1391 | 1610 |
| LMArena Russian | 1373 | 1537 |
| LMArena Spanish | 1358 | 1497 |
| LMArena French | 1385 | — |
| LMArena German | 1374 | — |
| LMArena Japanese | 1333 | — |
| LMArena Korean | 1319 | — |
Instruction Following Gemini 4 Argon leads
DeepSeek-V3: 72.8 (#130), Gemini 4 Argon: 80.1 (#2)
| Benchmark | DeepSeek-V3 | Gemini 4 Argon |
|---|---|---|
| LMArena Instruction Following | 1345 | 1538 |
| LiveBench Instruction Following | 81.5% | — |
| IFEval | 83.2% | — |
Long Context Gemini 4 Argon leads
DeepSeek-V3: 34.0 (#253), Gemini 4 Argon: 47.6 (#15)
| Benchmark | DeepSeek-V3 | Gemini 4 Argon |
|---|---|---|
| LMArena Longer Query | 1352 | 1549 |
| Fiction.LiveBench | 50% | — |
Writing & Preference Gemini 4 Argon leads
DeepSeek-V3: 57.4 (#130), Gemini 4 Argon: 71.4 (#19)
| Benchmark | DeepSeek-V3 | Gemini 4 Argon |
|---|---|---|
| LMArena Text | 1375 | 1534 |
| LMArena Creative Writing | 1364 | 1531 |
| LMArena Multi-Turn | 1389 | 1557 |
| Short-Story Creative Writing | 77% | — |
| EQ-Bench Creative Writing | 1472 | — |
| WildBench | 83% | — |
| LiveBench Language | 49.1% | — |
Frequently asked questions
Is DeepSeek-V3 better than Gemini 4 Argon?
Gemini 4 Argon is the stronger model overall, scoring 56.5 to 39.5 on the Noometry Index.
Is DeepSeek-V3 or Gemini 4 Argon better for coding?
Gemini 4 Argon scores higher on coding benchmarks: 62.0 versus 42.3 in the Noometry coding category.
How many benchmarks do DeepSeek-V3 and Gemini 4 Argon share?
15 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Gemini 4 Argon has 21.