Compare / Gemini 3.1 Pro Preview vs GPT-6 Astra
Gemini 3.1 Pro Preview vs GPT-6 Astra
64 results from 6 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 | GPT-6 Astra |
|---|---|---|---|
| BulletBenchmeasured by us | Cost per gameUS dollars | $0.21 | $0.28 |
| BulletBenchmeasured by us | Games lost on time% of games | 50.0% | 37.5% |
| BulletBenchmeasured by us | Invalid moves% of moves | 0.0% | 0.0% |
| BulletBenchmeasured by us | Ladder Eloladder Elo | 683 | 839 |
| BulletBenchmeasured by us | Median move timemilliseconds | 6.2 s | 5.6 s |
| BulletBenchmeasured by us | Cost per gameUS dollars | $0.0446 | $0.0943 |
| BulletBenchmeasured by us | Games lost on time% of games | 58.3% | 75.0% |
| BulletBenchmeasured by us | Invalid moves% of moves | 0.0% | 0.0% |
| BulletBenchmeasured by us | Ladder Eloladder Elo | 529 | 323 |
| BulletBenchmeasured by us | Median move timemilliseconds | 3.1 s | 2.3 s |
| CatalogBenchmeasured by us | Channel rules broken% of products | 4.8% | 0.0% |
| CatalogBenchmeasured by us | Content quality% of checks | 97.2% | 99.0% |
| CatalogBenchmeasured by us | Failed outputs% of products | 0.0% | 0.0% |
| CatalogBenchmeasured by us | Missing UK information% of products | 0.0% | 0.0% |
| CatalogBenchmeasured by us | Not findable% of products | 6.0% | 5.4% |
| CatalogBenchmeasured by us | Publish-ready listings% of products | 27.4% | 72.6% |
| CatalogBenchmeasured by us | Reliably publish-ready% of products | 10.7% | 67.9% |
| CatalogBenchmeasured by us | Unsupported claimsclaims per product | 2.67 | 0.07 |
| CatalogBenchmeasured by us | Unsupported claims% of products | 70.8% | 3.0% |
| CatalogBenchmeasured by us | Wrong attributes% of products | 16.1% | 20.2% |
| CatalogBenchmeasured by us | Wrong category or variant% of products | 1.8% | 1.8% |
| CatalogBenchmeasured by us | Channel compliance% of products | 98.8% | 100.0% |
| CatalogBenchmeasured by us | Channel rules broken% of products | 1.2% | 0.0% |
| CatalogBenchmeasured by us | Conflicts caught% of conflicts | 100.0% | 100.0% |
| CatalogBenchmeasured by us | Content quality% of checks | 96.8% | 98.3% |
| CatalogBenchmeasured by us | Cost per productUS dollars | $0.0355 | $0.0371 |
| CatalogBenchmeasured by us | Decision accuracy% of decisions | 96.4% | 98.4% |
| CatalogBenchmeasured by us | Failed outputs% of products | 0.0% | 0.0% |
| CatalogBenchmeasured by us | Field accuracy% of missing fields | 95.4% | 94.2% |
| CatalogBenchmeasured by us | Invented values% of filled values | 1.5% | 1.9% |
| CatalogBenchmeasured by us | Missing UK information% of products | 0.0% | 0.0% |
| CatalogBenchmeasured by us | Not findable% of products | 7.7% | 6.0% |
| CatalogBenchmeasured by us | Publish-ready listings% of products | 58.3% | 77.4% |
| CatalogBenchmeasured by us | Reliably publish-ready% of products | 41.1% | 75.0% |
| CatalogBenchmeasured by us | Unsupported claims% of products | 26.8% | 1.2% |
| CatalogBenchmeasured by us | Unsupported claimsclaims per product | 0.68 | 0.02 |
| CatalogBenchmeasured by us | Wrong attributes% of products | 14.9% | 17.3% |
| CatalogBenchmeasured by us | Wrong category or variant% of products | 1.8% | 1.8% |
| ROASBenchmeasured by us | Audience scorescore out of 100 | 67.5 | 68.6 |
| ROASBenchmeasured by us | Business scorescore out of 100 | 35.9 | 57.5 |
| ROASBenchmeasured by us | Consistency scorescore out of 100 | 33.9 | 44.7 |
| ROASBenchmeasured by us | Contribution profitsimulated US dollars | $507,736 | $599,139 |
| ROASBenchmeasured by us | Cost of a runUS dollars | $0.52 | $1.23 |
| ROASBenchmeasured by us | Months over budgetmonths of 12 | 0 | 0 |
| ROASBenchmeasured by us | Overall scorescore out of 100 | 43.5 | 56.6 |
| ROASBenchmeasured by us | Planning scorescore out of 100 | 54.9 | 54.9 |
| ROASBenchmeasured by us | Return on ad spendprofit per $1 spent | $0.42 | $0.67 |
| Arena (formerly LMArena)reported | Confirmed task successIPS effect estimate | -0.08 | 0.13 |
| Arena (formerly LMArena)reported | Praise over complaintIPS effect estimate | -0.01 | 0.34 |
| Arena (formerly LMArena)reported | SteerabilityIPS effect estimate | -0.04 | -0.01 |
| Arena (formerly LMArena)reported | Tool groundingIPS effect estimate | -0.00 | 0.00 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,459 | 1,473 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,472 | 1,465 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,477 | 1,483 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,480 | 1,453 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,509 | 1,510 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,479 | 1,473 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,487 | 1,478 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,480 | 1,469 |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 19.0% | 0.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.34 | 0.38 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 72.2 | 68.9 |
| Vectara Hallucination Leaderboardreported | Answer rate% of documents | 99.4% | 100.0% |
| Vectara Hallucination Leaderboardreported | Hallucination rate% of summaries | 10.4% | 8.7% |
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.