Join us in making real-work benchmark results accessible across industries. Contribute a model and/or benchmark to the Specialized Intelligence Index.
- You bring the eval. Tasks and data in Harbor-compatible format.
- We run it, standardized. Diversity, calibration, and contamination checks, then one harness across the full roster at no cost.
- You stay in control. We publish scores only, with your approval. Your eval is never republished or redistributed unless you choose.
- Models are eligible too. Fine-tuned models are evaluated under the same terms.
