Compare / DeepSeek V4 Pro 0423 vs Gemini 3.1 Pro Preview
DeepSeek V4 Pro 0423 vs Gemini 3.1 Pro Preview
31 results from 5 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | DeepSeek V4 Pro 0423 | Gemini 3.1 Pro Preview |
|---|---|---|---|
| ROASBenchmeasured by us | Audience scorescore out of 100 | 44.8 | 67.5 |
| ROASBenchmeasured by us | Business scorescore out of 100 | 23.8 | 35.9 |
| ROASBenchmeasured by us | Consistency scorescore out of 100 | 9.1 | 33.9 |
| ROASBenchmeasured by us | Contribution profitsimulated US dollars | $301,289 | $507,736 |
| ROASBenchmeasured by us | Cost of a runUS dollars | $0.20 | $0.52 |
| ROASBenchmeasured by us | Months over budgetmonths of 12 | 0 | 0 |
| ROASBenchmeasured by us | Overall scorescore out of 100 | 28.2 | 43.5 |
| ROASBenchmeasured by us | Planning scorescore out of 100 | 54.0 | 54.9 |
| ROASBenchmeasured by us | Return on ad spendprofit per $1 spent | $0.26 | $0.42 |
| Arena (formerly LMArena)reported | Confirmed task successIPS effect estimate | -0.04 | -0.08 |
| Arena (formerly LMArena)reported | Praise over complaintIPS effect estimate | -0.02 | -0.01 |
| Arena (formerly LMArena)reported | SteerabilityIPS effect estimate | -0.03 | -0.04 |
| Arena (formerly LMArena)reported | Tool groundingIPS effect estimate | -0.00 | -0.00 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,455 | 1,472 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,460 | 1,477 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,446 | 1,480 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,480 | 1,509 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,452 | 1,479 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,458 | 1,487 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,451 | 1,480 |
| OpenHands Indexreported | Average% resolved | 40.7% | 60.6% |
| OpenHands Indexreported | Front end (SWE-Bench Multimodal)% resolved | 36.8% | 44.1% |
| OpenHands Indexreported | Greenfield (Commit0)% resolved | 12.5% | 25.0% |
| OpenHands Indexreported | Information gathering (GAIA)% resolved | 12.7% | 81.8% |
| OpenHands Indexreported | Issue resolution (SWE-Bench)% resolved | 73.2% | 76.8% |
| OpenHands Indexreported | Testing (SWT-Bench)% resolved | 68.1% | 75.1% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 34.0% | 19.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.36 | 0.34 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 68.4 | 72.2 |
| Vectara Hallucination Leaderboardreported | Answer rate% of documents | 97.2% | 99.4% |
| Vectara Hallucination Leaderboardreported | Hallucination rate% of summaries | 8.6% | 10.4% |
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.