RISE Humanities Data Benchmark, 0.5.5-pre1

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

Your search for Benchmark 'magazine_pages__true' with Search Hidden 'False' returned 87 results, showing page 3 of 9.
Result 21 of 87

Test T1447 at 2026-07-25

newspaper-page document-understanding printed latin 20 en prose, columns

Configuration
Providergenai
Modelgemini-3.5-flash-lite
  
Temperature0.0
DataclassMagazinePage
  
Normalized Score0.00 %
Test time13.13 seconds
Prompt 

Extract all advertisements and return their bounding boxes.
The original size of the page is {width} x {height} pixels.

Results 

{"advertisements":[{"box":[36.0,36.0,965.0,964.0]}]}

Scoring
Fuzzy Score F1 micro / macro Micro precision/recall Tue/False Positives
n/a n/a n/a n/a n/a n/a n/a n/a n/a
      Micro Precision Micro Recall Instances TP FP FN
Costs / Pricing
Pricing Date: n/a, n/a. Tokens: 52.3K IT + 4.8K OT = 57.1K TTCost: 0.016$ + 0.012$ = 0.028$
Cite: Hindermann, Marti, Alberto, et al., (2025). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.16941752
Result 22 of 87

Test T1357 at 2026-07-25

newspaper-page document-understanding printed latin 20 en prose, columns

Configuration
Provideropenai
Modelgpt-5.6-sol
  
Temperature1.0
DataclassMagazinePage
  
Normalized Score94.70 %
Test time5.73 seconds
Prompt 

Extract all advertisements and return their bounding boxes.
The original size of the page is {width} x {height} pixels.

Results 

{"advertisements":[{"box":[1383.0,1814.0,2311.0,2503.0]},{"box":[334.0,2475.0,2306.0,3183.0]}]}

Scoring
Fuzzy Score F1 micro / macro Micro precision/recall Tue/False Positives
n/a n/a n/a n/a n/a n/a n/a n/a n/a
      Micro Precision Micro Recall Instances TP FP FN
Costs / Pricing
Pricing Date: n/a, n/a. Tokens: 158.1K IT + 14.2K OT = 172.3K TTCost: 0.791$ + 0.426$ = 1.216$
Cite: Hindermann, Marti, Alberto, et al., (2025). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.16941752
Result 23 of 87

Test T1202 at 2026-07-02

newspaper-page document-understanding printed latin 20 en prose, columns

Configuration
Provideranthropic
Modelclaude-fable-5
  
Temperature0.0
DataclassMagazinePage
  
Normalized Score96.00 %
Test time6.85 seconds
Prompt 

Extract all advertisements and return their bounding boxes.
The original size of the page is {width} x {height} pixels.

Results 

{"advertisements":[{"box":[265.0,210.0,2330.0,915.0]},{"box":[265.0,975.0,1280.0,1690.0]},{"box":[1330.0,975.0,2330.0,1690.0]},{"box":[265.0,1780.0,1280.0,2480.0]},{"box":[1330.0,1780.0,2330.0,2480.0]},{"box":[265.0,2570.0,1280.0,3230.0]},{"box":[1330.0,2570.0,2330.0,3230.0]}]}

Scoring
Fuzzy Score F1 micro / macro Micro precision/recall Tue/False Positives
n/a n/a n/a n/a n/a n/a n/a n/a n/a
      Micro Precision Micro Recall Instances TP FP FN
Costs / Pricing
Pricing Date: n/a, n/a. Tokens: 213.6K IT + 4.2K OT = 217.8K TTCost: 2.136$ + 0.210$ = 2.346$
Cite: Hindermann, Marti, Alberto, et al., (2025). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.16941752
Result 24 of 87

Test T1189 at 2026-07-01

newspaper-page document-understanding printed latin 20 en prose, columns

Configuration
Provideranthropic
Modelclaude-sonnet-5
  
Temperature0.0
DataclassMagazinePage
  
Normalized Score4.80 %
Test time3.21 seconds
Prompt 

Extract all advertisements and return their bounding boxes.
The original size of the page is {width} x {height} pixels.

Results 

{"advertisements":[{"box":[178.0,95.0,2361.0,990.0]},{"box":[178.0,1032.0,2321.0,2261.0]}]}

Scoring
Fuzzy Score F1 micro / macro Micro precision/recall Tue/False Positives
n/a n/a n/a n/a n/a n/a n/a n/a n/a
      Micro Precision Micro Recall Instances TP FP FN
Costs / Pricing
Pricing Date: n/a, n/a. Tokens: 216.6K IT + 4.0K OT = 220.5K TTCost: 0.433$ + 0.040$ = 0.473$
Cite: Hindermann, Marti, Alberto, et al., (2025). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.16941752
Result 25 of 87

Test T1176 at 2026-06-29

newspaper-page document-understanding printed latin 20 en prose, columns

Configuration
Providergenai
Modelgemini-3.1-flash-lite
  
Temperature0.0
DataclassMagazinePage
  
Normalized Score0.00 %
Test time1.87 seconds
Prompt 

Extract all advertisements and return their bounding boxes.
The original size of the page is {width} x {height} pixels.

Results 

{"advertisements":[{"box":[131.0,83.0,924.0,472.0]},{"box":[127.0,483.0,924.0,579.0]},{"box":[127.0,588.0,514.0,683.0]},{"box":[537.0,588.0,924.0,683.0]},{"box":[127.0,692.0,924.0,884.0]}]}

Scoring
Fuzzy Score F1 micro / macro Micro precision/recall Tue/False Positives
n/a n/a n/a n/a n/a n/a n/a n/a n/a
      Micro Precision Micro Recall Instances TP FP FN
Costs / Pricing
Pricing Date: n/a, n/a. Tokens: 52.3K IT + 5.0K OT = 57.4K TTCost: 0.013$ + 0.008$ = 0.021$
Cite: Hindermann, Marti, Alberto, et al., (2025). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.16941752
Result 26 of 87

Test T1163 at 2026-06-22

newspaper-page document-understanding printed latin 20 en prose, columns

Configuration
Providerscicore
Modelqwen35-397b-a17b-fp8
  
Temperature0.0
DataclassMagazinePage
  
Normalized Score8.70 %
Test time13.17 seconds
Prompt 

Extract all advertisements and return their bounding boxes.
The original size of the page is {width} x {height} pixels.

Results 

{"advertisements":[{"box":[76.0,73.0,882.0,913.0]}]}

Scoring
Fuzzy Score F1 micro / macro Micro precision/recall Tue/False Positives
n/a n/a n/a n/a n/a n/a n/a n/a n/a
      Micro Precision Micro Recall Instances TP FP FN
Costs / Pricing
Pricing Date: n/a, n/a. Tokens: 135.3K IT + 253.3K OT = 388.7K TTCost: 0.000$ + 0.000$ = 0.000$
Cite: Hindermann, Marti, Alberto, et al., (2025). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.16941752
Result 27 of 87

Test T1114 at 2026-06-08

newspaper-page document-understanding printed latin 20 en prose, columns

Configuration
Providerx-ai
Modelgrok-4.3
  
Temperature0.0
DataclassMagazinePage
  
Normalized Score61.80 %
Test time13.89 seconds
Prompt 

Extract all advertisements and return their bounding boxes.
The original size of the page is {width} x {height} pixels.

Results 

{"advertisements":[{"box":[150.0,120.0,2330.0,3150.0]}]}

Scoring
Fuzzy Score F1 micro / macro Micro precision/recall Tue/False Positives
n/a n/a n/a n/a n/a n/a n/a n/a n/a
      Micro Precision Micro Recall Instances TP FP FN
Costs / Pricing
Pricing Date: n/a, n/a. Tokens: 125.3K IT + 3.0K OT = 128.3K TTCost: 0.157$ + 0.008$ = 0.164$
Cite: Hindermann, Marti, Alberto, et al., (2025). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.16941752
Result 28 of 87

Test T1127 at 2026-06-05

newspaper-page document-understanding printed latin 20 en prose, columns

Configuration
Provideropenrouter
Modelmeta-llama/llama-4-scout-17b-16e-instruct
  
Temperature0.0
DataclassMagazinePage
  
Normalized Score0.00 %
Test time174.80 seconds
Prompt 

Extract all advertisements and return their bounding boxes.
The original size of the page is {width} x {height} pixels.

Results 

{"advertisements":[{"box":[93.0,51.0,492.0,463.0]},{"box":[93.0,493.0,492.0,804.0]},{"box":[506.0,493.0,911.0,802.0]}]}

Scoring
Fuzzy Score F1 micro / macro Micro precision/recall Tue/False Positives
n/a n/a n/a n/a n/a n/a n/a n/a n/a
      Micro Precision Micro Recall Instances TP FP FN
Costs / Pricing
Pricing Date: n/a, n/a. Tokens: 109.8K IT + 4.7K OT = 114.5K TTCost: 0.009$ + 0.001$ = 0.010$
Cite: Hindermann, Marti, Alberto, et al., (2025). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.16941752
Result 29 of 87

Test T1140 at 2026-06-05

newspaper-page document-understanding printed latin 20 en prose, columns

Configuration
Provideropenrouter
Modelstepfun/step-3.7-flash-20260528
  
Temperature0.0
DataclassMagazinePage
  
Normalized Score12.00 %
Test time522.51 seconds
Prompt 

Extract all advertisements and return their bounding boxes.
The original size of the page is {width} x {height} pixels.

Results 

{"advertisements":[{"box":[110.0,637.0,867.0,744.0]}]}

Scoring
Fuzzy Score F1 micro / macro Micro precision/recall Tue/False Positives
n/a n/a n/a n/a n/a n/a n/a n/a n/a
      Micro Precision Micro Recall Instances TP FP FN
Costs / Pricing
Pricing Date: n/a, n/a. Tokens: 57.0K IT + 211.1K OT = 268.1K TTCost: 0.011$ + 0.243$ = 0.254$
Cite: Hindermann, Marti, Alberto, et al., (2025). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.16941752
Result 30 of 87

Test T1088 at 2026-06-04

newspaper-page document-understanding printed latin 20 en prose, columns

Configuration
Provideranthropic
Modelclaude-opus-4-8
  
Temperature0.0
DataclassMagazinePage
  
Normalized Score81.60 %
Test time4.59 seconds
Prompt 

Extract all advertisements and return their bounding boxes.
The original size of the page is {width} x {height} pixels.

Results 

{"advertisements":[{"box":[343.0,343.0,2287.0,1716.0]},{"box":[347.0,1761.0,1310.0,2110.0]},{"box":[1525.0,1761.0,2287.0,2110.0]},{"box":[347.0,2150.0,1310.0,2580.0]},{"box":[1360.0,2150.0,2287.0,2580.0]},{"box":[330.0,2620.0,2305.0,3225.0]}]}

Scoring
Fuzzy Score F1 micro / macro Micro precision/recall Tue/False Positives
n/a n/a n/a n/a n/a n/a n/a n/a n/a
      Micro Precision Micro Recall Instances TP FP FN
Costs / Pricing
Pricing Date: n/a, n/a. Tokens: 213.6K IT + 4.2K OT = 217.8K TTCost: 1.068$ + 0.105$ = 1.173$
Cite: Hindermann, Marti, Alberto, et al., (2025). RISE-UNIBAS/humanities_data_benchmark, 10.5281/zenodo.16941752