# benchmarks.bio results: choose a version V1 and V2 are different evaluation sets and are not a like-for-like time series. There is no mixed-version overall leaderboard. V2 is the newer available result set, but V1 covers more historical models. The interactive site still defaults to V1 for Overall/Capabilities and to V2 for BioSecBench-Refusal. ## V1 — broader coverage Broader historical model coverage. The default interactive Overall/Capabilities view uses V1. Available benchmark scores are equally weighted; missing scores are excluded rather than treated as zero. The static overall table currently has 23 model-and-harness rows. V1 data updated: 2026-09-20. - [V1 HTML](https://benchmarks.bio/results/v1/) - [V1 JSON](https://benchmarks.bio/results/v1/latest.json) - [V1 CSV](https://benchmarks.bio/results/v1/latest.csv) - [V1 Markdown](https://benchmarks.bio/results/v1/index.md) ## V2 — newer, narrower coverage Newer clean-v1 reruns and updated evaluation sets, currently tested on fewer frontier models. The V2 overall view requires complete coverage and uses configured evaluation-count weights. BioSecBench-Refusal defaults to V2 in the interactive site. The static overall table currently has 9 complete-coverage model-and-harness rows. V2 data updated: 2026-09-21. Source/provenance: V2 score arrays in src/App.tsx, checked against the per-evaluation files in public/data/v2/. BioSecBench-Refusal scores come from public/data/v2/refusalbench.json. Overall weights come from public/data/aggregate-leaderboard.json. The V2 page explains its score weights and lists the V2 evaluation counts. - [V2 HTML](https://benchmarks.bio/results/v2/) - [V2 JSON](https://benchmarks.bio/results/v2/latest.json) - [V2 CSV](https://benchmarks.bio/results/v2/latest.csv) - [V2 Markdown](https://benchmarks.bio/results/v2/index.md) For programmatic discovery, read [versions.json](https://benchmarks.bio/results/versions.json). For the interactive comparison, visit [benchmarks.bio](https://benchmarks.bio/).