RISE Humanities Data Benchmark, 0.6.1

Search Test Runs

 

A test run is a single execution of a benchmark test using a defined model configuration.
Each run represents how a particular large language model (LLM) — such as GPT-4, Claude-3, or Gemini — performed on a given task at a specific time, with specific settings.

A test run includes:

  • Prompt and role definition – what the model was asked to do and from what perspective (e.g. “as a historian”).
  • Model configuration – provider, model version, temperature, and other generation parameters.
  • Results – the model’s actual response and its evaluation (scores such as F1 or accuracy).
  • Usage and cost data – token counts and calculated API costs.
  • Metadata – information like the test date, benchmark name, and person who executed it.

Together, test runs make it possible to compare models, providers, and configurations across benchmarks in a transparent and reproducible way.

Search Results
Show compact results Refine Search New Search
Download JSON Download CSV (Downloads are limited to 250 results.)

Your search for Benchmark 'business_letters__true' with Search Hidden 'False' returned 613 results, showing page 1 of 62.
Result 1 of 613

Test T1872 at 2026-09-29

letter information-extraction typed, handwritten 20 de prose person, location

anthropic claude-sonnet-5-5 · temp 0.0 · dataclass Document

Response vs ground truth

One model response from this run. The per-input comparison needs JavaScript.

Receiver persons fn0
Receiver persons fp0
Receiver persons tp1
Send date fn0
Send date fp0
Send date tp1
Sender persons fn0
Sender persons fp0
Sender persons tp0

{"metadata":{"send_date":["1926-03-29"],"letter_title":null,"sender_persons":null,"receiver_persons":["Dr. W. Krasting"]}}

Scoring

F1 micro62.0%
F1 macro52.0%

Speed

Model time, all inputs66.5 s
Mean per input2.66 s
Slowest input7.10 s
Inputs timed25

Summed model time, not elapsed: requests run in parallel and the worker count is not recorded, so the run finished sooner than this.

Costs / Pricing

Input tokens164.5K
Output tokens3.3K
Total tokens167.8K
Input cost$0.329
Output cost$0.0331
Total cost$0.3621

Priced from 2026-09-28 · Input $/M $2.00 · Output $/M $10.00

Cite: Hindermann, Marti, Alberto, et al., (2026). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.21975827
Result 2 of 613

Test T1873 at 2026-09-29

letter information-extraction typed, handwritten 20 de prose person, location

anthropic claude-sonnet-5-5 · temp 0.0 · dataclass Document

Response vs ground truth

One model response from this run. The per-input comparison needs JavaScript.

Receiver persons fn0
Receiver persons fp0
Receiver persons tp1
Send date fn0
Send date fp0
Send date tp1
Sender persons fn2
Sender persons fp1
Sender persons tp0

{"metadata":{"send_date":["1926-02-16"],"letter_title":null,"sender_persons":["Basler Rheinschiffahrt-Aktiengesellschaft"],"receiver_persons":["Herr Christ"]}}

Scoring

F1 micro58.0%
F1 macro63.0%

Speed

Model time, all inputs92.7 s
Mean per input2.90 s
Slowest input6.37 s
Inputs timed32

Summed model time, not elapsed: requests run in parallel and the worker count is not recorded, so the run finished sooner than this.

Costs / Pricing

Input tokens280.7K
Output tokens4.4K
Total tokens285.1K
Input cost$0.5613
Output cost$0.044
Total cost$0.6054

Priced from 2026-09-28 · Input $/M $2.00 · Output $/M $10.00

Cite: Hindermann, Marti, Alberto, et al., (2026). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.21975827
Result 3 of 613

Test T1676 at 2026-09-29

letter information-extraction typed, handwritten 20 de prose person, location

anthropic claude-fable-5-1 · temp 0.0 · dataclass Document

Response vs ground truth

One model response from this run. The per-input comparison needs JavaScript.

Receiver persons fn0
Receiver persons fp0
Receiver persons tp1
Send date fn0
Send date fp0
Send date tp1
Sender persons fn2
Sender persons fp1
Sender persons tp0

{"metadata":{"send_date":["1926-02-16"],"letter_title":null,"sender_persons":["Basler Rheinschiffahrt-Aktiengesellschaft"],"receiver_persons":["Christ"]}}

Scoring

F1 micro57.0%
F1 macro56.0%

Speed

Model time, all inputs534.2 s
Mean per input9.37 s
Slowest input15.98 s
Inputs timed57

Summed model time, not elapsed: requests run in parallel and the worker count is not recorded, so the run finished sooner than this.

Costs / Pricing

Input tokens445.2K
Output tokens23.3K
Total tokens468.4K
Input cost$4.45
Output cost$1.16
Total cost$5.62

Priced from 2026-09-08 · Input $/M $10.00 · Output $/M $50.00

Cite: Hindermann, Marti, Alberto, et al., (2026). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.21975827
Result 4 of 613

Test T1796 at 2026-09-29

letter information-extraction typed, handwritten 20 de prose person, location

scicore Qwen3.8-Flash-Next-FP8 · temp 0.0 · dataclass Document

Response vs ground truth

One model response from this run. The per-input comparison needs JavaScript.

Receiver persons fn0
Receiver persons fp0
Receiver persons tp1
Send date fn0
Send date fp0
Send date tp1
Sender persons fn2
Sender persons fp1
Sender persons tp0

{"metadata":{"send_date":["1926-02-16"],"letter_title":null,"sender_persons":["Basler Rheinschiffahrt-Aktiengesellschaft"],"receiver_persons":["Herr Christ"]}}

Scoring

F1 micro53.0%
F1 macro52.0%

Speed

Model time, all inputs1596.4 s
Mean per input28.01 s
Slowest input154.42 s
Inputs timed57

Summed model time, not elapsed: requests run in parallel and the worker count is not recorded, so the run finished sooner than this.

Costs / Pricing

Input tokens304.9K
Output tokens169.7K
Total tokens474.6K
Input cost$0.00
Output cost$0.00
Total cost$0.00

Priced from 2026-09-27 · Input $/M $0.00 · Output $/M $0.00

Cite: Hindermann, Marti, Alberto, et al., (2026). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.21975827
Result 5 of 613

Test T1871 at 2026-09-29

letter information-extraction typed, handwritten 20 de prose person, location

anthropic claude-sonnet-5-5 · temp 0.0 · dataclass Document

Response vs ground truth

One model response from this run. The per-input comparison needs JavaScript.

Receiver persons fn0
Receiver persons fp0
Receiver persons tp1
Send date fn0
Send date fp0
Send date tp1
Sender persons fn2
Sender persons fp1
Sender persons tp0

{"metadata":{"send_date":["1926-02-16"],"letter_title":null,"sender_persons":["Basler Rheinschiffahrt-Aktiengesellschaft"],"receiver_persons":["Herr Christ"]}}

Scoring

F1 micro59.0%
F1 macro59.0%

Speed

Model time, all inputs176.8 s
Mean per input3.10 s
Slowest input6.06 s
Inputs timed57

Summed model time, not elapsed: requests run in parallel and the worker count is not recorded, so the run finished sooner than this.

Costs / Pricing

Input tokens445.2K
Output tokens7.7K
Total tokens452.9K
Input cost$0.8903
Output cost$0.077
Total cost$0.9674

Priced from 2026-09-28 · Input $/M $2.00 · Output $/M $10.00

Cite: Hindermann, Marti, Alberto, et al., (2026). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.21975827
Result 6 of 613

Test T1811 at 2026-09-29

letter information-extraction typed, handwritten 20 de prose person, location

anthropic claude-opus-5-5 · temp 0.0 · dataclass Document

Response vs ground truth

One model response from this run. The per-input comparison needs JavaScript.

Receiver persons fn0
Receiver persons fp0
Receiver persons tp1
Send date fn0
Send date fp0
Send date tp1
Sender persons fn2
Sender persons fp1
Sender persons tp0

{"metadata":{"send_date":["1926-02-16"],"letter_title":null,"sender_persons":["F. Ritter"],"receiver_persons":["Christ"]}}

Scoring

F1 micro76.0%
F1 macro73.0%

Speed

Model time, all inputs320.5 s
Mean per input5.62 s
Slowest input9.60 s
Inputs timed57

Summed model time, not elapsed: requests run in parallel and the worker count is not recorded, so the run finished sooner than this.

Costs / Pricing

Input tokens445.2K
Output tokens14.8K
Total tokens459.9K
Input cost$1.78
Output cost$0.2951
Total cost$2.08

Priced from 2026-09-27 · Input $/M $4.00 · Output $/M $20.00

Cite: Hindermann, Marti, Alberto, et al., (2026). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.21975827
Result 7 of 613

Test T1677 at 2026-09-29

letter information-extraction typed, handwritten 20 de prose person, location

anthropic claude-fable-5-1 · temp 0.0 · dataclass Document

Response vs ground truth

One model response from this run. The per-input comparison needs JavaScript.

Receiver persons fn0
Receiver persons fp0
Receiver persons tp1
Send date fn0
Send date fp0
Send date tp1
Sender persons fn0
Sender persons fp0
Sender persons tp0

{"metadata":{"send_date":["1926-03-29"],"letter_title":null,"sender_persons":null,"receiver_persons":["Dr. W. Krasting"]}}

Scoring

F1 micro67.0%
F1 macro55.0%

Speed

Model time, all inputs210.6 s
Mean per input8.42 s
Slowest input12.17 s
Inputs timed25

Summed model time, not elapsed: requests run in parallel and the worker count is not recorded, so the run finished sooner than this.

Costs / Pricing

Input tokens164.5K
Output tokens9.7K
Total tokens174.2K
Input cost$1.65
Output cost$0.4831
Total cost$2.13

Priced from 2026-09-08 · Input $/M $10.00 · Output $/M $50.00

Cite: Hindermann, Marti, Alberto, et al., (2026). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.21975827
Result 8 of 613

Test T1797 at 2026-09-29

letter information-extraction typed, handwritten 20 de prose person, location

scicore Qwen3.8-Flash-Next-FP8 · temp 0.0 · dataclass Document

Response vs ground truth

One model response from this run. The per-input comparison needs JavaScript.

Receiver persons fn0
Receiver persons fp0
Receiver persons tp1
Send date fn0
Send date fp0
Send date tp1
Sender persons fn0
Sender persons fp0
Sender persons tp0

{"metadata":{"send_date":["1926-03-29"],"letter_title":["null"],"sender_persons":["null"],"receiver_persons":["Dr. W. Krasting"]}}

Scoring

F1 micro60.0%
F1 macro52.0%

Speed

Model time, all inputs461.3 s
Mean per input18.45 s
Slowest input56.14 s
Inputs timed25

Summed model time, not elapsed: requests run in parallel and the worker count is not recorded, so the run finished sooner than this.

Costs / Pricing

Input tokens110.3K
Output tokens64.7K
Total tokens175K
Input cost$0.00
Output cost$0.00
Total cost$0.00

Priced from 2026-09-27 · Input $/M $0.00 · Output $/M $0.00

Cite: Hindermann, Marti, Alberto, et al., (2026). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.21975827
Result 9 of 613

Test T1678 at 2026-09-29

letter information-extraction typed, handwritten 20 de prose person, location

anthropic claude-fable-5-1 · temp 0.0 · dataclass Document

Response vs ground truth

One model response from this run. The per-input comparison needs JavaScript.

Receiver persons fn0
Receiver persons fp0
Receiver persons tp1
Send date fn0
Send date fp0
Send date tp1
Sender persons fn2
Sender persons fp2
Sender persons tp0

{"metadata":{"send_date":["1926-02-16"],"letter_title":null,"sender_persons":["Basler Rheinschiffahrt-Aktiengesellschaft","F. Ritter"],"receiver_persons":["Christ"]}}

Scoring

F1 micro58.0%
F1 macro63.0%

Speed

Model time, all inputs315.6 s
Mean per input9.86 s
Slowest input17.83 s
Inputs timed32

Summed model time, not elapsed: requests run in parallel and the worker count is not recorded, so the run finished sooner than this.

Costs / Pricing

Input tokens280.7K
Output tokens14.9K
Total tokens295.6K
Input cost$2.81
Output cost$0.7457
Total cost$3.55

Priced from 2026-09-08 · Input $/M $10.00 · Output $/M $50.00

Cite: Hindermann, Marti, Alberto, et al., (2026). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.21975827
Result 10 of 613

Test T1798 at 2026-09-29

letter information-extraction typed, handwritten 20 de prose person, location

scicore Qwen3.8-Flash-Next-FP8 · temp 0.0 · dataclass Document

Response vs ground truth

One model response from this run. The per-input comparison needs JavaScript.

Receiver persons fn0
Receiver persons fp0
Receiver persons tp1
Send date fn0
Send date fp0
Send date tp1
Sender persons fn2
Sender persons fp1
Sender persons tp0

{"metadata":{"send_date":["1926-02-16"],"letter_title":null,"sender_persons":["Basler Rheinschiffahrt-Aktiengesellschaft"],"receiver_persons":["Christ"]}}

Scoring

F1 micro51.0%
F1 macro56.0%

Speed

Model time, all inputs642.7 s
Mean per input20.09 s
Slowest input65.45 s
Inputs timed32

Summed model time, not elapsed: requests run in parallel and the worker count is not recorded, so the run finished sooner than this.

Costs / Pricing

Input tokens193.7K
Output tokens89.7K
Total tokens283.4K
Input cost$0.00
Output cost$0.00
Total cost$0.00

Priced from 2026-09-27 · Input $/M $0.00 · Output $/M $0.00

Cite: Hindermann, Marti, Alberto, et al., (2026). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.21975827