University of Basel RISE · Research and Infrastructure SupportHumanities Data Benchmark

gemini-3.8-flash

Google
average±2.1 (95%)
85.6
leaderboard rankof 114
≈8
benchmarksof 12
12
per 1,000 documentsmean over its benchmarks
$27
seconds per inputmean response time
23.7
Compare with other models Its row in the Matrix All its test runs

By task

  • Data correction

    71.7

    Best of 114 models with all its benchmarks: claude-opus-5-5, 82.7.

  • Document understanding

    90.7

    Best of 107 models with all its benchmarks: gpt-6-astra, 97.0.

  • Information extraction

    89.0

    Best of 107 models with all its benchmarks: grok-4.6, 90.8.

  • Transcription

    92.0

    Best of 114 models with all its benchmarks: gemini-4-argon, 96.2.

By benchmark

BenchmarkScoreThe best current run on each benchmark, on the benchmark's metric scaled to 0–100. ± is the 95% sampling margin, from resampling the run's inputs.RankRank among all current models on that benchmark; ≈ for models that cannot be told apart within sampling error.Per 1,000 docss / inputBest run
Bibliographic Data 94.5±2.5 ≈57of 115 $53$0.26 run 42.81 T1645 · 3 Sep 2026prompt.txt
Blacklist Cards 96.5±1.9 ≈14of 114 $4.77$0.16 run 5.32 T1652 · 16 Sep 2026prompt.txt
Book Advert XML Repair 46.0±10.3 ≈16of 119 $34$1.69 run 29.26 T1655 · 3 Sep 2026prompt.txt
Business Letters 74.0±7.2 ≈1of 114 $5.96$0.34 run 6.51 T1646 · 16 Sep 2026prompt.txt
Company Lists 79.9±12.9 ≈58of 114 $15$0.23 run 14.95 T1653 · 16 Sep 2026prompt.txt
Duty Rosters 92.2±2.7 ≈10of 107 $76$1.59 run 62.93 T1659 · 3 Sep 2026prompt.txt
Fraktur Adverts 96.8±3.6 ≈27of 114 $46$0.23 run 38.84 T1649 · 3 Sep 2026prompt_optimized.txt
General Meeting Minutes 95.6±2.7 ≈1of 108 $32$0.29 run 26.59 T1658 · 3 Sep 2026prompt.txt
Library Cards 90.0±1.4 ≈1of 114 $6.04$1.59 run 5.59 T1650 · 16 Sep 2026prompt.txt
Magazine Pages 82.6±10.5 ≈18of 115 $6.67$0.31 run 7.92 T1657 · 3 Sep 2026prompt.txt
Medieval Manuscripts 82.0±12.9 ≈1of 114 $29$0.35 run 30.5 T1651 · 3 Sep 2026prompt.txt
Personnel Cards 97.3±1.0 ≈1of 114 $16$0.99 run 13.28 T1656 · 16 Sep 2026prompt.txt

Price

FromInput, $ per million tokensOutput, $ per million tokensSource
2 Sep 20260.753.75archived price page

Every run

  1. T1645Bibliographic Data26 Sep 202670.1$0.21
  2. T1659Duty Rosters16 Sep 202691.9$1.69
  3. T1658General Meeting Minutes16 Sep 202695.4$0.22
  4. T1657Magazine Pages16 Sep 202677.4$0.24
  5. T1656Personnel Cards16 Sep 202697.3$0.99
  6. T1655Book Advert XML Repair16 Sep 202644.9$1.67
  7. T1654Company Lists16 Sep 202664.4$0.23
  8. T1653Company Lists16 Sep 202679.9$0.23
  9. T1652Blacklist Cards16 Sep 202696.5$0.16
  10. T1651Medieval Manuscripts16 Sep 202680.1$0.54
  11. T1650Library Cards16 Sep 202690.0$1.59
  12. T1649Fraktur Adverts16 Sep 202677.0$0.27
  13. T1648Business Letters16 Sep 2026 · 32 signed letters70.0$0.26
  14. T1647Business Letters16 Sep 2026 · 25 unsigned letters94.0$0.096
  15. T1646Business Letters16 Sep 2026 · all 57 letters74.0$0.34
  16. T1645Bibliographic Data16 Sep 202673.3$0.26
  17. T1659Duty Rosters3 Sep 202692.2$1.59
  18. T1658General Meeting Minutes3 Sep 202695.6$0.29
  19. T1657Magazine Pages3 Sep 202682.6$0.31
  20. T1656Personnel Cards3 Sep 202696.7$0.95
  21. T1655Book Advert XML Repair3 Sep 202646.0$1.69
  22. T1654Company Lists3 Sep 202665.7$0.21
  23. T1653Company Lists3 Sep 202676.5$0.22
  24. T1652Blacklist Cards3 Sep 202694.4$0.18
  25. T1651Medieval Manuscripts3 Sep 202682.0$0.35
  26. T1650Library Cards3 Sep 202689.8$1.65
  27. T1649Fraktur Adverts3 Sep 202696.8$0.23
  28. T1648Business Letters3 Sep 2026 · 32 signed letters71.0$0.21
  29. T1647Business Letters3 Sep 2026 · 25 unsigned letters94.0$0.11
  30. T1646Business Letters3 Sep 2026 · all 57 letters73.0$0.37
  31. T1645Bibliographic Data3 Sep 202694.5$0.26