
Microsoft-Decision-1
Microsoft-Decision-1 is Microsoft's first decision-only AI: instead of writing text, it returns a yes/no answer, a choice, or a score along with a probability. Where the company's MAI models draw pictures or transcribe speech, this one handles sorting requests, sending them to the right place, grading AI answers, and deciding whether an agent's next action should be blocked. Built by retraining a small Alibaba Qwen model, on our own Korean and Japanese judgment tests it landed at about the same level as Jev and Perplexity Decider, with answers that barely changed when the options were shuffled. It is weaker on exam-style knowledge questions and on picking the better of two answers, and its reported confidence matches its real accuracy less closely than its rivals'. It reads text only, long documents are hard to fit in one request, and because the company keeps updating what it serves, results under the same name can change over time.
API cost
Expert Performance Evaluation
Classification and judgment
All classification and judgment AI rankingsSee each evaluation’s detail page for its source and last updateData sources & removal requests