An empty cell is an absence of recorded evidence
Examenos displays an em dash when it has no recorded result for a model and metric. A source may not list the model, a result may not have been published, or it may not yet have been verified for this dataset. The gap alone does not tell us which explanation applies.
Coverage describes the dataset
Data coverage counts model/metric pairs with at least one recorded result. Several effort or harness records for the same pair count once. Official and independent results have separate denominators because they belong to different source families.
A model with more documented benchmarks is not automatically more capable. A model with an independent evaluator gap did not necessarily fail that evaluator's tests. Treat missing cells as limits on the comparison you can make.
Choose comparisons with shared evidence
The Compare tool normally hides single-model benchmark rows when several models are selected. You can reveal those rows, but they still cannot establish a direct comparison without another model's score under a relevant setup.
What you can do
Open Data coverage, select a model, and switch between missing and recorded results. Browse a model profile for the available evidence, or inspect the benchmark results directory.
If you have a primary source for a missing result, use the site's Feedback button to suggest it. Suggestions are reviewed; a submission does not directly change scores. See our sourcing and correction policy.
Sources and further reading
These explanations describe Examenos's data and display policy. Numeric examples are available in the linked model/benchmark records, each with original citations. Examenos curates results rather than conducting these evaluations.