What it measures

Vals combines the two exam years and averages pass@1 across eight runs per model. This is a mixed-year cohort, separate from the AIME 2024 and AIME 2025 articles.

How to read the result

A mean across repeated runs helps reduce sampling noise on a small test. Match the question years, sampling settings and answer extraction before comparing with other contest scores.

Published evaluation

The linked publication documents the task and scoring procedure. Model results retain the source’s evaluation setting, native score units and available uncertainty.

Where this fits in an evaluation suite

Final-answer accuracy, proof credit, and expert-reviewed research progress are distinct signals. A proof-aware evaluation should retain partial work and audit strong claims.

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

Sixty questions spanning AIME 2024 and 2025; eight runs per model; source-defined answer extraction and averaged pass@1.

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.

PRIMARY EVIDENCE & FURTHER READING

  1. AIME — evaluation methodology and results ↗

    Vals AI · 2026

Reviewed 2026-10-26. Explanation and interpretation by RLTasks. Benchmark ownership remains with its original authors.