Compare / DeepSeek V3.2 vs Gemini 3.8 Flash
DeepSeek V3.2 vs Gemini 3.8 Flash
16 results from 3 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | DeepSeek V3.2 | Gemini 3.8 Flash |
|---|---|---|---|
| Arena (formerly LMArena)reported | OverallArena rating | 1,437 | 1,477 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,423 | 1,482 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,400 | 1,482 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,453 | 1,529 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,420 | 1,488 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,425 | 1,492 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,410 | 1,486 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 44.2 | 76.3 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 21.5 | 40.9 |
| Artificial Analysisreported | GPQA Diamond% of questions | 84.0% | 95.3% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 24.6% | 47.8% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 73.3% | 84.0% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 46.8% | 87.6% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 26.0% | 13.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.36 | 0.34 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 57.3 | 78.5 |
Bold green marks the better value on that metric. Where a source reports ranges that overlap, the difference may not be meaningful; see the benchmark page for ranges.