Codesota · FinderTasks · benchmarks · evidenceNo account required
Finder · Interactive

Find the benchmark that actually fits.

Describe the model or product you are evaluating. Codesota narrows it to the closest research area, task page, and benchmark path.

01Start broadUse product language first; the finder will translate it into benchmark taxonomy.
02Refine quicklyPick a domain, then select one or more concrete tasks.
03Open resultsJump straight to browse pages, or send us a missing benchmark request.
Benchmark FinderStep 1 of 4
Describe the work

What are you trying to evaluate?

Paste the product requirement, model capability, benchmark question, or a rough sentence.