OmevraOmevra · Research
Research systems
Open research built alongside the platform. Each system is live, and each publishes its own measurements — including the results that land below baseline.
Web Agent
An autonomous agent that operates ordinary web interfaces — forms, tables, dialogs, portals — checking each step and recovering when one fails. In its final evaluation it completed 36 of 36 tasks with its recovery mechanisms on (95% CI 90.3–100%) and 27 of 36 with them off (95% CI 57.8–87.9%); on the adversarial suite, 9 of 9 on against 0 of 9 off.
Hybrid Perturb-Bench
A benchmark for predicting how single cells respond to genetic perturbations, on the published Norman 2019 Perturb-seq dataset. The hybrid singles model scored 0.537, below the global-mean baseline of 0.588, against a measured ceiling of 0.917 — below baseline, published as-is.
Multi-Agent Evidence
Three cooperating agents — one retrieves the evidence, one drafts the summary, one checks every citation. Measured head to head against a single agent, both systems scored 1.00 on citation accuracy — the ceiling of that benchmark, so the test cannot separate them — and the multi-agent system cost more to get there.