University of Basel RISE · Research and Infrastructure SupportHumanities Data Benchmark

gpt-6-astra

OpenAI
average±1.6 (95%)
87.5
leaderboard rankof 114
≈1
benchmarksof 12
12
per 1,000 documentsmean over its benchmarks
$100
seconds per inputmean response time
22.9
Compare with other models Its row in the Matrix All its test runs

By task

  • Data correction

    78.3

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

  • Document understanding

    97.0

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

  • Information extraction

    87.1

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

  • Transcription

    94.9

    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 99.5±0.4 ≈1of 115 $197$0.98 run 32.44 T1660 · 4 Sep 2026prompt.txt
Blacklist Cards 96.2±2.4 ≈14of 114 $42$1.39 run 4.69 T1667 · 4 Sep 2026prompt.txt
Book Advert XML Repair 58.8±10.8 ≈1of 119 $42$2.12 run 18.59 T1670 · 16 Sep 2026prompt.txt
Business Letters 62.0±6.9 ≈24of 114 $75$4.27 run 5.59 T1661 · 4 Sep 2026prompt.txt
Company Lists 69.9±12.7 ≈84of 114 $90$1.36 run 18.9 T1668 · 16 Sep 2026prompt.txt
Duty Rosters 96.3±1.3 ≈1of 107 $127$2.67 run 35.83 T1674 · 16 Sep 2026prompt.txt
Fraktur Adverts 97.0±1.1 ≈15of 114 $204$1.02 run 49.84 T1664 · 4 Sep 2026prompt_optimized.txt
General Meeting Minutes 94.2±3.2 ≈1of 108 $101$0.91 run 18.09 T1673 · 4 Sep 2026prompt.txt
Library Cards 91.5±1.2 ≈1of 114 $30$8.00 run 7.32 T1665 · 17 Sep 2026prompt.txt
Magazine Pages 96.8±2.8 ≈1of 115 $44$2.02 run 4.39 T1672 · 4 Sep 2026prompt.txt
Medieval Manuscripts 89.9±5.8 ≈1of 114 $142$1.70 run 67.38 T1666 · 16 Sep 2026prompt.txt
Personnel Cards 97.8±0.8 ≈1of 114 $104$6.37 run 11.44 T1671 · 4 Sep 2026prompt.txt

Price

FromInput, $ per million tokensOutput, $ per million tokensSource
3 Sep 202610.050.0archived price page

Every run

  1. T1665Library Cards17 Sep 202691.5$8.00
  2. T1674Duty Rosters16 Sep 202696.3$2.67
  3. T1673General Meeting Minutes16 Sep 202692.5$0.87
  4. T1672Magazine Pages16 Sep 202696.8$2.03
  5. T1671Personnel Cards16 Sep 202697.7$6.35
  6. T1670Book Advert XML Repair16 Sep 202658.8$2.12
  7. T1669Company Lists16 Sep 202668.5$1.23
  8. T1668Company Lists16 Sep 202669.9$1.36
  9. T1667Blacklist Cards16 Sep 202695.4$1.40
  10. T1666Medieval Manuscripts16 Sep 202689.9$1.70
  11. T1664Fraktur Adverts16 Sep 202696.7$0.98
  12. T1663Business Letters16 Sep 2026 · 32 signed letters66.0$2.72
  13. T1662Business Letters16 Sep 2026 · 25 unsigned letters59.0$1.51
  14. T1661Business Letters16 Sep 2026 · all 57 letters62.0$4.26
  15. T1660Bibliographic Data16 Sep 202699.5$0.98
  16. T1674Duty Rosters4 Sep 202696.1$2.59
  17. T1673General Meeting Minutes4 Sep 202694.2$0.91
  18. T1672Magazine Pages4 Sep 202696.8$2.02
  19. T1671Personnel Cards4 Sep 202697.8$6.37
  20. T1670Book Advert XML Repair4 Sep 202655.0$2.11
  21. T1669Company Lists4 Sep 202668.3$1.23
  22. T1668Company Lists4 Sep 202669.5$1.34
  23. T1667Blacklist Cards4 Sep 202696.2$1.39
  24. T1666Medieval Manuscripts4 Sep 202683.6$1.66
  25. T1665Library Cards4 Sep 202691.1$8.09
  26. T1664Fraktur Adverts4 Sep 202697.0$1.02
  27. T1663Business Letters4 Sep 2026 · 32 signed letters65.0$2.72
  28. T1662Business Letters4 Sep 2026 · 25 unsigned letters62.0$1.51
  29. T1661Business Letters4 Sep 2026 · all 57 letters62.0$4.27
  30. T1660Bibliographic Data4 Sep 202699.5$0.98