University of Basel RISE · Research and Infrastructure SupportHumanities Data Benchmark

gpt-4o

OpenAI
average
66.4
leaderboard rankof 114
≈77
benchmarksof 12
12
per 1,000 documentsmean over its benchmarks
$9.90
seconds per inputmean response time
71.6
Compare with other models Its row in the Matrix All its test runs

By task

  • Data correction

    53.2

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

  • Document understanding

    67.7

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

  • Information extraction

    77.8

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

  • Transcription

    65.5

    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 Dataolder run 92.0 ≈72of 115 –– run 24.55 T0007 · 2 Mar 2025prompt.txt
Blacklist Cardsolder run 93.9 ≈45of 114 $4.05$0.13 run 5.65 T0305 · 24 Oct 2025prompt.txt
Book Advert XML Repair 15.8±6.4 ≈89of 119 $6.98$0.35 run 11.97 T0446 · 25 Jan 2026prompt.txt
Business Lettersolder run 53.0 ≈90of 114 –– run 3.99 T0010 · 5 Mar 2025prompt.txt
Company Lists 69.2±15.4 ≈84of 114 $9.74$0.15 run 24.81 T0337 · 26 Jan 2026prompt.txt
Duty Rosters 72.7±3.3 ≈64of 107 $20$0.41 run 30.03 T1296 · 24 Jul 2026prompt.txt
Fraktur Advertsolder run 41.4 96of 114 $16$0.079 run 103.29 T0079 · 30 Sep 2025prompt_optimized.txt
General Meeting Minutes 74.5±12.1 ≈85of 108 $8.62$0.078 run 12.75 T1247 · 3 Jul 2026prompt.txt
Library Cardsolder run 89.5 ≈15of 114 –– run 4.91 T0066 · 2 Sep 2025prompt.txt
Magazine Pages 39.8±12.8 ≈47of 115 $3.64$0.17 run 4.45 T0717 · 21 Mar 2026prompt.txt
Medieval Manuscripts 64.4±12.2 ≈72of 114 $6.31$0.076 run 611.94 T0275 · 26 Jan 2026prompt.txt
Personnel Cards 90.6±2.9 ≈78of 114 $14$0.87 run 21.21 T0608 · 10 Feb 2026prompt.txt

Price

FromInput, $ per million tokensOutput, $ per million tokensSource
0 Sep 20252.510.0archived price page

Every run

  1. T1296Duty Rosters16 Sep 202671.8$0.41
  2. T1247General Meeting Minutes16 Sep 202672.8$0.074
  3. T0717Magazine Pages16 Sep 202639.2$0.17
  4. T1296Duty Rosters24 Jul 202672.7$0.41
  5. T1247General Meeting Minutes3 Jul 202674.5$0.078
  6. T0717Magazine Pages21 Mar 202639.8$0.17
  7. T0608Personnel Cards10 Feb 202690.6$0.87
  8. T0608Personnel Cards9 Feb 202689.8$0.87
  9. T0337Company Lists26 Jan 202669.2$0.15
  10. T0305Blacklist Cards26 Jan 202693.5$0.16
  11. T0275Medieval Manuscripts26 Jan 202664.4$0.076
  12. T0446Book Advert XML Repair25 Jan 202615.8$0.35
  13. T0338Company Lists25 Jan 202657.0$0.13
  14. T0079Fraktur Adverts24 Jan 202637.3$0.23
  15. T0066Library Cards24 Jan 202685.1$1.46
  16. T0039Business Letters24 Jan 2026 · 32 signed letters57.0$0.23
  17. T0038Business Letters24 Jan 2026 · 25 unsigned letters45.0$0.14
  18. T0010Business Letters24 Jan 2026 · all 57 letters52.0$0.37
  19. T0007Bibliographic Data24 Jan 202689.3$0.12
  20. T0446Book Advert XML Repair8 Dec 202515.5$0.36
  21. T0338Company Lists28 Oct 202559.5$0.12
  22. T0337Company Lists28 Oct 202565.7$0.13
  23. T0305Blacklist Cards24 Oct 202593.9$0.13
  24. T0275Medieval Manuscripts24 Oct 202557.3$0.067
  25. T0079Fraktur Adverts30 Sep 202541.4$0.079
  26. T0066Library Cards30 Sep 202582.2$1.30
  27. T0039Business Letters30 Sep 2025 · 32 signed letters56.0$0.18
  28. T0038Business Letters30 Sep 2025 · 25 unsigned letters40.0$0.11
  29. T0010Business Letters30 Sep 2025 · all 57 letters52.0$0.28
  30. T0007Bibliographic Data30 Sep 202587.9$0.11
  31. T0066Library Cards2 Sep 202589.5–
  32. T0079Fraktur Adverts9 May 202538.6–
  33. T0079Fraktur Adverts7 May 20255.3–
  34. T0039Business Letters11 Apr 2025 · 32 signed letters56.0–
  35. T0038Business Letters11 Apr 2025 · 25 unsigned letters46.0–
  36. T0010Business Letters11 Apr 2025 · all 57 letters51.0–
  37. T0007Bibliographic Data8 Apr 202581.6–
  38. T0010Business Letters1 Apr 2025 · all 57 letters53.0–
  39. T0007Bibliographic Data1 Apr 202590.5–
  40. T0010Business Letters5 Mar 2025 · all 57 letters53.0–
  41. T0007Bibliographic Data5 Mar 202583.7–
  42. T0007Bibliographic Data4 Mar 202582.8–
  43. T0007Bibliographic Data2 Mar 202592.0–