Deepseek v4 Pro vs Gemini 3 Flash
Deepseek v4 Pro 
Gemini 3 Flash Verdict
- Deepseek v4 Pro (high) leads on quality: 60.4 vs 59.3.
- Deepseek v4 Pro (high) is stronger in coding, human preference, maths, reasoning.
- Gemini 3 Flash (thinking) is stronger in agents & tools, composite, instruction following, knowledge, long context, multimodal.
| Metric | Deepseek v4 Pro (high) | Gemini 3 Flash (thinking) |
|---|---|---|
| BenchLeader Index | 60.4 | 59.3 |
| Coding score | 59.3 | – |
| Human preference score | 66.0 | – |
| Maths score | 66.2 | – |
| Reasoning score | 65.9 | – |
| Agents & tools score | – | 62.6 |
| Composite score | – | 56.8 |
| Instruction following score | – | 72.4 |
| Knowledge score | – | 65.5 |
| Long context score | – | 62.9 |
| Multimodal score | – | 63.3 |
| Blended price $/M | – | $1.13 |
| Output speed | – | 214 tok/s |
| Time to first answer | – | 5.7 s |
| Context window | – | 1M |
| GPQA Diamond | 90.9% | – |
| OTIS Mock AIME | 95.6% | – |
| SciCode | 46.4% | – |
| WeirdML | 46.5% | – |
| LMArena Text | 1462 | – |
| LMArena Hard Prompts | 1482 | – |
| LMArena Coding | 1505 | – |
| LMArena WebDev | 1581 | – |
| AA Intelligence Index | – | 26.3 |
| IFBench | – | 78.0% |
| AA-LCR | – | 78.0% |
| MMMU-Pro | – | 79.9% |
| AA-Omniscience | – | 10.1 |
| Terminal-Bench Hard | – | 38.6% |
| GPQA Diamond (AA) | – | 89.8% |
| Humanity's Last Exam (AA) | – | 36.6% |
| τ²-Bench Telecom (AA) | – | 80.4% |
Data as of 2026-09-13. Best configuration of each model; every score links to its source on the model pages.
Deepseek v4 Pro vs Gemini 3 Flash: questions
- Is Deepseek v4 Pro better than Gemini 3 Flash?
- Deepseek v4 Pro (high) leads on quality: 60.4 vs 59.3. The BenchLeader Index combines every independent quality benchmark; Deepseek v4 Pro (high) is ahead overall as of 2026-09-13, but check the category scores for your use.