← Models

Model profile

Step 3.7 Flash

2026-05-23release date
2eval lineages
1discovery sources

Evidence summary

Published evidence spans 2 evals and 2 of 7 behavior components. Its strongest relative result is AA-Omniscience (hallucination_rate, #192 of 311); its weakest is TAC (base_welfare_rate, #56 of 68).

Compare this model

Only models sharing at least one published sub-eval are listed.

Official and reference links

Published eval results

Rank is within that sub-eval. Black marks the observed result; the grey dot marks the value implied by the global rank. Values stay on each source’s native scale.

Eval / sub-evalRankValueDistributionBetterSource
AA-Omnisciencehallucination_rate#192 / 3110.8438↓ lowerSource ↗official
TACbase_welfare_rate#56 / 6819.87↑ higherSource ↗self run