Compare / Claude Haiku 4.5 vs GPT-6 Sol
Claude Haiku 4.5 vs GPT-6 Sol
46 results from 5 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | Claude Haiku 4.5 | GPT-6 Sol |
|---|---|---|---|
| BulletBenchmeasured by us | Games lost on time% of games | 25.0% | 50.0% |
| BulletBenchmeasured by us | Ladder Eloladder Elo | 445 | 445 |
| BulletBenchmeasured by us | Median move timemilliseconds | 1.2 s | 1.6 s |
| BulletBenchmeasured by us | Return on ad spendprofit per $1 spent | $0.31 | $0.65 |
| CatalogBenchmeasured by us | Channel rules broken% of products | 39.3% | 0.6% |
| CatalogBenchmeasured by us | Content quality% of checks | 94.4% | 97.9% |
| 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 | 6.5% | 7.1% |
| CatalogBenchmeasured by us | Publish-ready listings% of products | 0.0% | 67.9% |
| CatalogBenchmeasured by us | Reliably publish-ready% of products | 0.0% | 55.4% |
| CatalogBenchmeasured by us | Unsupported claimsclaims per product | 14.3 | 0.36 |
| CatalogBenchmeasured by us | Unsupported claims% of products | 99.4% | 11.9% |
| CatalogBenchmeasured by us | Wrong attributes% of products | 51.2% | 13.7% |
| CatalogBenchmeasured by us | Wrong category or variant% of products | 6.5% | 3.0% |
| CatalogBenchmeasured by us | Channel rules broken% of products | 23.2% | 0.0% |
| CatalogBenchmeasured by us | Conflicts caught% of conflicts | 100.0% | 100.0% |
| CatalogBenchmeasured by us | Content quality% of checks | 94.6% | 97.7% |
| CatalogBenchmeasured by us | Cost per productUS dollars | $0.0061 | $0.0085 |
| CatalogBenchmeasured by us | Decision accuracy% of decisions | 86.0% | 97.5% |
| CatalogBenchmeasured by us | Failed outputs% of products | 0.0% | 0.0% |
| CatalogBenchmeasured by us | Field accuracy% of missing fields | 93.7% | 93.7% |
| CatalogBenchmeasured by us | Invented values% of filled values | 6.2% | 1.0% |
| CatalogBenchmeasured by us | Missing UK information% of products | 0.0% | 0.0% |
| CatalogBenchmeasured by us | Not findable% of products | 8.9% | 7.7% |
| CatalogBenchmeasured by us | Publish-ready listings% of products | 0.6% | 70.2% |
| CatalogBenchmeasured by us | Reliably publish-ready% of products | 0.0% | 60.7% |
| CatalogBenchmeasured by us | Unsupported claimsclaims per product | 7.92 | 0.24 |
| CatalogBenchmeasured by us | Unsupported claims% of products | 94.0% | 7.1% |
| CatalogBenchmeasured by us | Wrong attributes% of products | 48.2% | 14.3% |
| CatalogBenchmeasured by us | Wrong category or variant% of products | 7.7% | 3.6% |
| ROASBenchmeasured by us | Audience scorescore out of 100 | 30.9 | 69.4 |
| ROASBenchmeasured by us | Business scorescore out of 100 | 7.3 | 52.9 |
| ROASBenchmeasured by us | Consistency scorescore out of 100 | 8.7 | 42.9 |
| ROASBenchmeasured by us | Contribution profitsimulated US dollars | −$205,001 | $600,357 |
| ROASBenchmeasured by us | Months over budgetmonths of 12 | 2 | 0 |
| ROASBenchmeasured by us | Overall scorescore out of 100 | 17.5 | 54.1 |
| ROASBenchmeasured by us | Planning scorescore out of 100 | 54.4 | 54.8 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,418 | 1,424 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,387 | 1,451 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,452 | 1,507 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,415 | 1,447 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,414 | 1,457 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,399 | 1,440 |
| Vectara Hallucination Leaderboardreported | Answer rate% of documents | 99.5% | 100.0% |
| Vectara Hallucination Leaderboardreported | Hallucination rate% of summaries | 9.8% | 6.5% |
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.