Model comparison
DeepSeek V4.1 Flash vs Gemini 3 Pro
Gemini 3 Pro is the stronger model overall, scoring 54.7 to 50.3 on the Noometry Index.
Last verified . 21 shared benchmarks.
DeepSeek
GoogleSummary
- They share 21 benchmarks with published results for both. DeepSeek V4.1 Flash scores higher in 3 categories and Gemini 3 Pro in 5 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Gemini 3 Pro leads 64.3 to 41.6.
- The biggest single-benchmark swing is ProofBench: 54% for DeepSeek V4.1 Flash and 20% for Gemini 3 Pro.
- DeepSeek V4.1 Flash has downloadable open weights; the other is API-only.
Side by side
| DeepSeek V4.1 Flash | Gemini 3 Pro | |
|---|---|---|
| Provider | DeepSeek | |
| Noometry Index | 50.3 | 54.7 |
| Released | 2026-09-09 | 2025-11-18 |
| Weights | Open | Proprietary |
| Context window | 1M | — |
| Max output | 393K | — |
| Input $ / M tokens | $0.15 | — |
| Output $ / M tokens | $0.60 | — |
| Results tracked | 25 | 67 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
DeepSeek V4.1 Flash: 52.8 (#29), Gemini 3 Pro: 52.1 (#33)
| Benchmark | DeepSeek V4.1 Flash | Gemini 3 Pro |
|---|---|---|
| LMArena WebDev | 1620 | 1439 |
| LMArena Coding | 1511 | 1482 |
| ALE-Bench | 1,092 | 1,177 |
| SWE-bench Verified | — | 72.9% |
| SWE-bench Verified (bash only) | — | 74.2% |
| SWE-bench Multilingual | — | 68.7% |
| SciCode | 51.9% | — |
| GSO | — | 18.6% |
| WeirdML | — | 69.9% |
| AlgoTune | — | 1.83 |
Agentic & Tool Use Not comparable
DeepSeek V4.1 Flash: —, Gemini 3 Pro: 40.5 (#21)
| Benchmark | DeepSeek V4.1 Flash | 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 V4.1 Flash: 46.2 (#39), Gemini 3 Pro: 52.5 (#28)
| Benchmark | DeepSeek V4.1 Flash | Gemini 3 Pro |
|---|---|---|
| NYT Connections (extended) | 89.6% | 94.4% |
| CritPt | 14.3% | 6.9% |
| LMArena Hard Prompts | 1485 | 1481 |
| Epoch Capabilities Index | 155.01 | 152.96 |
| ARC-AGI-2 | — | 31.1% |
| SimpleBench | — | 76.4% |
| Kagi LLM Benchmark | — | 80.1% |
| ARC-AGI-1 | — | 75% |
| Chess Puzzles | — | 31% |
| EnigmaEval | — | 18.2% |
| DTBench | 89.9% | — |
| LMCA | 47% | — |
| Surface Evolver Bench | 46.3% | — |
| ForecastBench | — | 61.2 |
Math Too close to call
DeepSeek V4.1 Flash: 48.4 (#52), Gemini 3 Pro: 48.8 (#50)
| Benchmark | DeepSeek V4.1 Flash | Gemini 3 Pro |
|---|---|---|
| ProofBench | 54% | 20% |
| LMArena Math | 1493 | 1475 |
| MathArena Final-Answer Competitions | — | 67% |
| OTIS Mock AIME 2024-2025 | — | 91.4% |
| Omni-MATH | — | 55.5% |
| FrontierMath (Feb 2025 set) | — | 37.6% |
| FrontierMath Tier 4 (v1) | — | 18.8% |
Knowledge Gemini 3 Pro leads
DeepSeek V4.1 Flash: 41.6 (#97), Gemini 3 Pro: 64.3 (#15)
| Benchmark | DeepSeek V4.1 Flash | Gemini 3 Pro |
|---|---|---|
| LMArena Expert | 1505 | 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 V4.1 Flash: —, Gemini 3 Pro: 56.9 (#1)
| Benchmark | DeepSeek V4.1 Flash | Gemini 3 Pro |
|---|---|---|
| LMArena Vision | — | 1305 |
| GeoBench | — | 84% |
| VPCT | — | 91% |
| LMArena Document | — | 1434 |
Multilingual Gemini 3 Pro leads
DeepSeek V4.1 Flash: 55.5 (#29), Gemini 3 Pro: 57.1 (#14)
| Benchmark | DeepSeek V4.1 Flash | Gemini 3 Pro |
|---|---|---|
| LMArena Non-English | 1452 | 1474 |
| LMArena Chinese | 1506 | 1522 |
| LMArena French | 1448 | 1492 |
| LMArena Russian | 1473 | 1493 |
| LMArena Spanish | 1499 | 1468 |
| LMArena German | — | 1517 |
| LMArena Japanese | — | 1513 |
| LMArena Korean | — | 1449 |
Instruction Following DeepSeek V4.1 Flash leads
DeepSeek V4.1 Flash: 77.6 (#24), Gemini 3 Pro: 76.4 (#42)
| Benchmark | DeepSeek V4.1 Flash | Gemini 3 Pro |
|---|---|---|
| LMArena Instruction Following | 1475 | 1458 |
| IFEval | — | 87.7% |
Long Context DeepSeek V4.1 Flash leads
DeepSeek V4.1 Flash: 45.2 (#46), Gemini 3 Pro: 44.0 (#75)
| Benchmark | DeepSeek V4.1 Flash | Gemini 3 Pro |
|---|---|---|
| LMArena Longer Query | 1476 | 1472 |
| CL-bench | — | 15.8% |
Writing & Preference Too close to call
DeepSeek V4.1 Flash: 65.8 (#43), Gemini 3 Pro: 66.6 (#34)
| Benchmark | DeepSeek V4.1 Flash | Gemini 3 Pro |
|---|---|---|
| LMArena Text | 1465 | 1479 |
| LMArena Creative Writing | 1436 | 1482 |
| EQ-Bench Creative Writing | 1540 | 1525 |
| LMArena Multi-Turn | 1460 | 1484 |
| WildBench | — | 85.9% |
Frequently asked questions
Is DeepSeek V4.1 Flash better than Gemini 3 Pro?
Gemini 3 Pro is the stronger model overall, scoring 54.7 to 50.3 on the Noometry Index.
Is DeepSeek V4.1 Flash or Gemini 3 Pro better for coding?
They score almost the same on coding (52.8 vs 52.1); test both on your own repository before choosing.
How many benchmarks do DeepSeek V4.1 Flash and Gemini 3 Pro share?
21 benchmarks have published results for both models. DeepSeek V4.1 Flash has 25 scored results on Noometry and Gemini 3 Pro has 67.