Professional questions require reasoning over long PDF documents and meeting task-specific requirements.
89 model results
One row per model and compatible score measure. Mean = equal-weight arithmetic average of imported reports. Distinct configurations reported for the same model in a source table remain separate. Reports can overlap; the range is not a confidence interval. Unbundle to inspect each observation.
31.00% · GDP.pdf · all-pass · Publisher setting: Max [1]
Published observations span 3.2 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
Leads this displayed comparison.
Published breakdowns
81.8% mean criterion credit; 31.0% of attempts satisfied every requirement.
31.00% · GDP.pdf · all-pass · Publisher setting: Max [1]
Published observations span 4 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#2 of 78 in this displayed comparison; 0.13 pp below the leader.
Published breakdowns
81.4% mean criterion credit; 31.0% of attempts satisfied every requirement.
32.60% · GDP.pdf · all-pass · Publisher setting: Max [1]
Published observations span 0.6 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#3 of 78 in this displayed comparison; 0.23 pp below the leader.
Published breakdowns
80.4% mean criterion credit; 32.6% of attempts satisfied every requirement.
30.60% · GDP.pdf · all-pass · Publisher setting: Max [1]
Published observations span 0.2 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#5 of 78 in this displayed comparison; 2.4 pp below the leader.
Published breakdowns
79.2% mean criterion credit; 30.6% of attempts satisfied every requirement.
Published observations span 4.4 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#11 of 78 in this displayed comparison; 4 pp below the leader.
Published breakdowns
81.4% mean criterion credit; 26.2% of attempts satisfied every requirement.
Published observations span 5.6 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#12 of 78 in this displayed comparison; 4.53 pp below the leader.
Published breakdowns
79.0% mean criterion credit; 25.8% of attempts satisfied every requirement.
Published observations span 1.4 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#17 of 78 in this displayed comparison; 6.23 pp below the leader.
Published breakdowns
79.7% mean criterion credit; 26.2% of attempts satisfied every requirement.
21.80% · GDP.pdf · all-pass · Publisher setting: High [1]
Published observations span 6.6 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#20 of 78 in this displayed comparison; 8.03 pp below the leader.
Published breakdowns
79.9% mean criterion credit; 21.8% of attempts satisfied every requirement.
22.80% · GDP.pdf · all-pass · Publisher setting: Max [1]
Published observations span 0.2 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#29 of 78 in this displayed comparison; 10.23 pp below the leader.
Published breakdowns
73.9% mean criterion credit; 22.8% of attempts satisfied every requirement.
Published observations span 2.8 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#34 of 78 in this displayed comparison; 11.73 pp below the leader.
Published breakdowns
70.2% mean criterion credit; 20.0% of attempts satisfied every requirement.
Publisher-selected recent-model cohort from the public evaluation page. Model labels preserve effort and fallback configurations. Values are a dated snapshot, not a live feed. Consult the publisher’s methodology for the harness, task split and judging; do not merge this cohort with vendor-run results.
Source-reported model settings are max unless a cell says otherwise. The original page reports numbers without a uniform percent symbol; original formatting is retained. OSWorld includes two scoring rules in each cell: leaderboard ordering uses partial credit only. The internal composite and AA index use separate scales; their scores cannot be averaged.
Epoch AI public data snapshot, CC BY 4.0. Externally collected result; original evaluation provider is linked. CSV: gdp_pdf_external.csv; score column: GDP.pdf score; fraction scale: 1. Model version is retained in configuration. Original source authors retain credit and any additional licensing terms.
Epoch AI public data snapshot, CC BY 4.0. Externally collected result; original evaluation provider is linked. CSV: gdp_pdf_external.csv; score column: GDP.pdf score; fraction scale: 1. Model version is retained in configuration. Original source authors retain credit and any additional licensing terms.
GDP.pdf · GDP.pdf w/ python · Kimi K3.1 Pro · complete release comparison · 2026-10-27
Published model comparison, 2026. Kimi K3.1 Pro · complete release comparison. Exact values from the publisher’s embedded full comparison table. Maximum effort, fallback settings and dated model versions remain in model labels. Blank cells stay missing. Source footnotes specify mixed harnesses and externally quoted comparisons; this is a launch-reported cohort rather than a new independent evaluation. Benchmarks · full embedded comparison table · row 16; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.
GDP.pdf · Kimi K3.1 Pro · complete release comparison · 2026-10-27
Published model comparison, 2026. Kimi K3.1 Pro · complete release comparison. Exact values from the publisher’s embedded full comparison table. Maximum effort, fallback settings and dated model versions remain in model labels. Blank cells stay missing. Source footnotes specify mixed harnesses and externally quoted comparisons; this is a launch-reported cohort rather than a new independent evaluation. Benchmarks · full embedded comparison table · row 36; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.
TRACE THE SCORE · Per-model notes use published breakdowns where available.
Score vs. release date
Epoch AI · 56 observations
GDP.pdf · GDP.pdf score · % · higher is better
Dates are model release dates recorded by Epoch AI, not evaluation dates. Each point retains its published score and configuration. Separate releases are not a matched longitudinal experiment. Missing dates are omitted. Explore all reported results ↓
The dashed line traces the best observed score through the model release dates. It describes this selection of reports, with their original configurations. It does not estimate unmeasured models or costs.
View plotted values and sources (56)
GDP.pdf · GDP.pdf · GDP.pdf score · %. Published measurements in their original score units.
These model reports include this benchmark. Their harnesses, task subsets, scoring policies, and resource budgets can differ; follow each report’s evaluation notes before joining the numbers.
Professional questions require reasoning over long PDF documents and meeting task-specific requirements.
How to read the result
Check document extraction and evidence synthesis separately; either can cause a correct-looking answer to fail.
Limits and comparison traps
Keep the document input route fixed. Native PDF, extracted text, and page images expose different information to a model.
Epoch data and evaluation setting
Epoch AI publishes this benchmark in its capability collection. Imported rows retain the published score column and model configuration. Original authors and evaluation providers remain credited in the result references.
Epoch attribution and score coverage
Data is attributed to Epoch AI under CC BY 4.0, with original evaluation providers credited in the linked source snapshots. Scores are imported only where a published mapping identifies the field and its scale; missing or incompatible records are omitted rather than assigned zero.
Published evaluation
Exact values from the publisher’s embedded full comparison table. Maximum effort, fallback settings and dated model versions remain in model labels. Blank cells stay missing. Source footnotes specify mixed harnesses and externally quoted comparisons; this is a launch-reported cohort rather than a new independent evaluation.
Where this fits in an evaluation suite
A large context window is a capacity specification. Successful retrieval and cross-document reasoning are observed capabilities. Test both at the actual lengths your workflow needs.
Start with a capability hypothesis, then choose a second test with a different failure mode. Keep the checkpoint, reasoning effort, agent tools, sampling policy, and scoring procedure fixed when measuring a training intervention. Record failures and missing outputs in the denominator.
A reproducible comparison
Exact values from the publisher’s embedded full comparison table. Maximum effort, fallback settings and dated model versions remain in model labels. Blank cells stay missing. Source footnotes specify mixed harnesses and externally quoted comparisons; this is a launch-reported cohort rather than a new independent evaluation.
Publish the exact run configuration, item coverage, score units, and available uncertainty. If a source supplies only an aggregate or a single run, preserve that limit. A new model name or a rounded score alone does not establish a capability improvement.