Compare / DeepSeek V4 Flash 0423 vs GPT-5.1
DeepSeek V4 Flash 0423 vs GPT-5.1
17 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 V4 Flash 0423 | GPT-5.1 |
|---|---|---|---|
| Arena (formerly LMArena)reported | OverallArena rating | 1,439 | 1,459 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,439 | 1,448 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,408 | 1,428 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,457 | 1,482 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,428 | 1,450 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,436 | 1,456 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,416 | 1,437 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 69.1 | 49.4 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 34.3 | 24.7 |
| Artificial Analysisreported | GPQA Diamond% of questions | 90.8% | 87.3% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 38.6% | 28.5% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 79.7% | 80.0% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 78.7% | 52.4% |
| Artificial Analysisreported | Τ-bench banking% of tasks | 39.4% | 15.9% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 12.0% | 113.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.36 | 0.38 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 54.6 | 55.8 |
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.