Compare / DeepSeek V3.2 Exp vs DeepSeek V4 Flash 0423
DeepSeek V3.2 Exp vs DeepSeek V4 Flash 0423
14 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 Exp | DeepSeek V4 Flash 0423 |
|---|---|---|---|
| Arena (formerly LMArena)reported | OverallArena rating | 1,446 | 1,439 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,422 | 1,439 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,409 | 1,408 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,449 | 1,457 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,416 | 1,428 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,425 | 1,436 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,410 | 1,416 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 16.6 | 34.3 |
| Artificial Analysisreported | GPQA Diamond% of questions | 79.7% | 90.8% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 14.9% | 38.6% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 72.3% | 79.7% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 25.0% | 12.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.35 | 0.36 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 54.0 | 54.6 |
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.