Compare / Gemini 3.1 Pro Preview vs Gemini 3.6 Flash
Gemini 3.1 Pro Preview vs Gemini 3.6 Flash
20 results from 3 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | Gemini 3.1 Pro Preview | Gemini 3.6 Flash |
|---|---|---|---|
| BulletBenchmeasured by us | Cost per gameUS dollars | $0.0446 | $0.0113 |
| BulletBenchmeasured by us | Games lost on time% of games | 58.3% | 41.7% |
| BulletBenchmeasured by us | Invalid moves% of moves | 0.0% | 0.3% |
| BulletBenchmeasured by us | Ladder Eloladder Elo | 529 | 628 |
| BulletBenchmeasured by us | Median move timemilliseconds | 3.1 s | 1.6 s |
| Arena (formerly LMArena)reported | Confirmed task successIPS effect estimate | -0.08 | -0.08 |
| Arena (formerly LMArena)reported | Praise over complaintIPS effect estimate | -0.01 | -0.09 |
| Arena (formerly LMArena)reported | SteerabilityIPS effect estimate | -0.04 | -0.04 |
| Arena (formerly LMArena)reported | Tool groundingIPS effect estimate | -0.00 | 0.00 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,459 | 1,453 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,472 | 1,470 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,477 | 1,472 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,480 | 1,469 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,509 | 1,500 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,479 | 1,473 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,487 | 1,482 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,480 | 1,472 |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 19.0% | 9.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.34 | 0.33 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 72.2 | 69.8 |
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.