Apps & data pipelines
Scira's open Amos 1 family spans small and large decision models for text and images, with answer shapes and object-box support depending on the variant.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
The Apache-2.0 family spans small and large models with a hosted demonstration. Its headline comparisons are author-run or submitted results, not independently verified rankings. The larger variants exceed some Jev-class cost limits, so size, serving hardware and benchmark rules matter.
Choose a model size and input type from the collection, then try a question with explicit allowed answers. If you compare its results with Jev, keep the test set and cost rules the same. A public-subset score is not a full benchmark result.