Model comparison
DeepSeek V4 Pro vs Gemini 3.8 Flash
Gemini 3.8 Flash is the stronger model overall, scoring 61.8 to 54.3 on the Noometry Index. DeepSeek V4 Pro costs 1.5× less per token, which makes it the better buy when Gemini 3.8 Flash's lead doesn't matter for your workload.
Last verified . 41 shared benchmarks.
DeepSeek
GoogleSummary
- They share 41 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 0 categories and Gemini 3.8 Flash in 9 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Gemini 3.8 Flash leads 76.9 to 56.5.
- The biggest single-benchmark swing is Surface Evolver Bench: 40% for DeepSeek V4 Pro and 76.9% for Gemini 3.8 Flash.
- DeepSeek V4 Pro is cheaper at $0.66 / $1.98 per million input/output tokens, against $0.75 / $3.75 for Gemini 3.8 Flash.
- Gemini 3.8 Flash accepts more context: 1.05M tokens versus 1M.
- DeepSeek V4 Pro has downloadable open weights; the other is API-only.
Side by side
| DeepSeek V4 Pro | Gemini 3.8 Flash | |
|---|---|---|
| Provider | DeepSeek | |
| Noometry Index | 54.3 | 61.8 |
| Released | 2026-04-24 | 2026-09-02 |
| Weights | Open | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 393K | 66K |
| Input $ / M tokens | $0.66 | $0.75 |
| Output $ / M tokens | $1.98 | $3.75 |
| Results tracked | 48 | 50 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemini 3.8 Flash leads
DeepSeek V4 Pro: 52.4 (#34), Gemini 3.8 Flash: 59.2 (#15)
| Benchmark | DeepSeek V4 Pro | Gemini 3.8 Flash |
|---|---|---|
| FrontierCode | 28.6% | 41.2% |
| LMArena WebDev | 1582 | 1584 |
| SciCode | 51% | 56.6% |
| WeirdML | 66.2% | 84.8% |
| LMArena Coding | 1470 | 1510 |
| ALE-Bench | 1,403 | 1,270 |
| SWE-bench Verified | 77.6% | — |
| DeepSWE | — | 73.8% |
| CursorBench | — | 39.6% |
| FrontierSWE | — | 19.6% |
Agentic & Tool Use Gemini 3.8 Flash leads
DeepSeek V4 Pro: 32.8 (#58), Gemini 3.8 Flash: 41.8 (#21)
| Benchmark | DeepSeek V4 Pro | Gemini 3.8 Flash |
|---|---|---|
| APEX-Agents | 47.3% | 64.3% |
| Vending-Bench 2 | 3,285 | 5,094 |
| Remote Labor Index | — | 5.8% |
| GDP.pdf | — | 23.4% |
Reasoning Gemini 3.8 Flash leads
DeepSeek V4 Pro: 56.5 (#24), Gemini 3.8 Flash: 76.9 (#5)
| Benchmark | DeepSeek V4 Pro | Gemini 3.8 Flash |
|---|---|---|
| ARC-AGI-2 | 61.3% | 89.2% |
| NYT Connections (extended) | 91.3% | 97.4% |
| ARC-AGI-1 | 90.5% | 98.5% |
| CritPt | 18% | 18.3% |
| Chess Puzzles | 47% | 61% |
| LMArena Hard Prompts | 1461 | 1508 |
| Mystery Game Puzzles | 43% | 47% |
| DTBench | 93.9% | 95.7% |
| LMCA | 45.5% | 52.9% |
| Surface Evolver Bench | 40% | 76.9% |
| Epoch Capabilities Index | 155.31 | 156.71 |
| Kagi LLM Benchmark | 53.5% | — |
| ForecastBench | 56.1 | — |
Math Too close to call
DeepSeek V4 Pro: 64.8 (#30), Gemini 3.8 Flash: 65.3 (#28)
| Benchmark | DeepSeek V4 Pro | Gemini 3.8 Flash |
|---|---|---|
| FrontierMath (Tiers 1-3) | 64.6% | 68.4% |
| FrontierMath Tier 4 | 26.8% | 22% |
| OTIS Mock AIME 2024-2025 | 98.6% | 98.9% |
| ProofBench | 50% | 48% |
| LMArena Math | 1455 | 1528 |
| MathArena Final-Answer Competitions | 76.6% | — |
Knowledge Gemini 3.8 Flash leads
DeepSeek V4 Pro: 59.5 (#31), Gemini 3.8 Flash: 74.8 (#2)
| Benchmark | DeepSeek V4 Pro | Gemini 3.8 Flash |
|---|---|---|
| GPQA Diamond | 91.7% | 95.4% |
| SimpleQA Verified | 52.9% | 69.7% |
| LMArena Expert | 1464 | 1524 |
| Humanity's Last Exam | — | 44.5% |
| Vectara Hallucination Rate | 8.6% | — |
Multimodal Not comparable
DeepSeek V4 Pro: —, Gemini 3.8 Flash: 40.7 (#45)
| Benchmark | DeepSeek V4 Pro | Gemini 3.8 Flash |
|---|---|---|
| LMArena Vision | — | 1314 |
| Blueprint-Bench 2 | — | 38.6% |
| Furniture Assembly | — | 31.7% |
Multilingual Gemini 3.8 Flash leads
DeepSeek V4 Pro: 54.4 (#45), Gemini 3.8 Flash: 58.0 (#5)
| Benchmark | DeepSeek V4 Pro | Gemini 3.8 Flash |
|---|---|---|
| LMArena Non-English | 1439 | 1491 |
| LMArena Chinese | 1486 | 1554 |
| LMArena French | 1472 | 1498 |
| LMArena German | 1458 | 1493 |
| LMArena Japanese | 1445 | 1502 |
| LMArena Korean | 1447 | 1459 |
| LMArena Russian | 1453 | 1515 |
| LMArena Spanish | 1458 | 1485 |
Instruction Following Gemini 3.8 Flash leads
DeepSeek V4 Pro: 76.1 (#47), Gemini 3.8 Flash: 78.0 (#13)
| Benchmark | DeepSeek V4 Pro | Gemini 3.8 Flash |
|---|---|---|
| LMArena Instruction Following | 1448 | 1490 |
Long Context Gemini 3.8 Flash leads
DeepSeek V4 Pro: 45.0 (#51), Gemini 3.8 Flash: 46.3 (#24)
| Benchmark | DeepSeek V4 Pro | Gemini 3.8 Flash |
|---|---|---|
| LMArena Longer Query | 1458 | 1508 |
| CL-bench Life | 13.5% | — |
Writing & Preference Gemini 3.8 Flash leads
DeepSeek V4 Pro: 65.5 (#46), Gemini 3.8 Flash: 72.2 (#15)
| Benchmark | DeepSeek V4 Pro | Gemini 3.8 Flash |
|---|---|---|
| LMArena Text | 1451 | 1499 |
| LMArena Creative Writing | 1446 | 1492 |
| EQ-Bench Creative Writing | 1553 | 1748 |
| LMArena Multi-Turn | 1467 | 1501 |
| EQ-Bench 4 | 1166 | — |
Frequently asked questions
Is DeepSeek V4 Pro better than Gemini 3.8 Flash?
Gemini 3.8 Flash is the stronger model overall, scoring 61.8 to 54.3 on the Noometry Index. DeepSeek V4 Pro costs 1.5× less per token, which makes it the better buy when Gemini 3.8 Flash's lead doesn't matter for your workload.
Which is cheaper, DeepSeek V4 Pro or Gemini 3.8 Flash?
DeepSeek V4 Pro is cheaper. It lists at $0.66 per million input tokens and $1.98 per million output tokens; Gemini 3.8 Flash lists at $0.75 and $3.75.
Is DeepSeek V4 Pro or Gemini 3.8 Flash better for coding?
Gemini 3.8 Flash scores higher on coding benchmarks: 59.2 versus 52.4 in the Noometry coding category.
Which has the bigger context window?
Gemini 3.8 Flash does, with 1.05M tokens against 1M.
How many benchmarks do DeepSeek V4 Pro and Gemini 3.8 Flash share?
41 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and Gemini 3.8 Flash has 50.