Every team asks which open model to use. Tryouts runs the same research questions through every model on Nebius Token Factory with Tavily search, scores each answer, and picks the cheapest model you can trust for your workload.
| Model | Correct | Wrong | $ per question |
|---|
One row per model, one column per question. Shaded columns are questions where the models disagree; those are the ones that tell models apart. Select a mark to replay that question.
The same question, two models, every search round and source.