PUBLIC DEMOS

Try a benchmark.

Choose models, run a fixed evaluation, and inspect the evidence. No signup required.

No accountNo API keyVisible grading

DEMO 01NO SIGNUP

MATH REASONING

GSM8K

Compare up to three models on a fixed set of multi-step grade-school math problems.

DEMO 02NO SIGNUP

DOMAIN KNOWLEDGE

Test technical skateboard knowledge and a precise product recommendation contract.

DEMO 03NO SIGNUP

SCIENCE REASONING

Compare models on challenging multiple-choice grade-school science questions.

DEMO 04NO SIGNUP

TRUTHFULNESS

Test whether models choose truthful answers over familiar and persuasive misconceptions.

DEMO 05NO SIGNUP

MEDICAL AI SAFETY

Inspect how models handle escalation, uncertainty, medication caution, and false reassurance.

YOUR EVALUATIONS