University of Basel RISE · Research and Infrastructure SupportHumanities Data Benchmark

Leaderboard

Which model for me?

Describe your documents and your project. The benchmarks closest to them are matched, and a model is recommended from its best runs there.

1 · My documents and task look like…Pick what your documents are. They are matched to the closest benchmarks.
Document type
Writing
Language
Task
Layout
Entry type
Century
No benchmarks selected yet.
2 · How many documents?Sets how much score is traded for efficiency at your scale: none for a one-off, up to 9 points for a pipeline.
Optimise for
3 · How sensitive is your data?Non-sensitive data may use cloud models; sensitive data must stay on premise (local models only). Always follow your institution's data-classification directive.

Your recommendation appears here

Pick what your documents are in question 1. The closest benchmarks are chosen for you, and you get the model whose best runs there fit your scale best, the models close to it, and what each costs per 1,000 documents.

Or start from a task:

To see every model on every benchmark instead, open the Matrix.