Assistant Benchmark compares assistants through recorded attempts at everyday tasks, giving readers evidence about what a product actually completed. Its October 8, 2026 snapshot reports 32 tested assistants within a catalog of 139. The scoring method assigns ratings from 1 to 10 against written criteria and requires a logged run before awarding a score. It distinguishes untested capabilities from inapplicable ones and keeps subjective personality judgments in a separate public-opinion category. An untested entry therefore represents missing evidence, not demonstrated failure.
Terms in this piece · Glossary
benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
