A Transcriptomic Benchmark for Foundation Models in Immunology and Inflammation
Our first multitask benchmark evaluates patient representations under the realistic constraints of immunology and inflammation cohorts.
Read the publicationPRIMO
Open collaboration to bring clarity to evaluations of foundation models for omics, from target discovery to the clinic.
Built by an open community of omics researchers, ML scientists, clinicians, and drug developers.
Updates
Latest from PRIMO
Public benchmark and leaderboard
The PRIMO tasks, datasets, baselines, and live leaderboard will be released on Hugging Face.
Hugging FaceOpen evaluation code
The evaluation harness, baseline implementations, and contribution workflow will be published on GitHub.
GitHubCommunity Contributors
PRIMO is initiated by Scienta Lab and is being built as an open, community-driven benchmark. We are welcoming founding contributors from academia, biotech, pharma, and technology.
Get involved
Contribute to PRIMO
Bring your expertise, datasets, models, or evaluation ideas. We welcome contributions from researchers and practitioners across the drug-development ecosystem.
Join the Community
Help build consensus on how omics foundation models should be evaluated.
Become a partnerShape the Benchmark
Contribute tasks, patient cohorts, modalities, endpoints, or methodological review.
Propose a contributionBenchmark a Model
Be among the first to compare your model with peers and strong statistical baselines.
Register interest