Benchmarks & research
Microsoft's own decision model, a hosted version of Qwen3.5-9B retrained to put a calibrated probability on every answer option in one pass.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
It handles yes/no, multiple-choice and rating questions in Microsoft Foundry, with OpenRouter listed as coming soon. Its price, $0.042 per million input tokens with free output, matches Jev's. Microsoft's model card rules out text generation and using it as the only decision-maker about people's credit, jobs, housing or health.
Use it for repeated small calls with fixed answers, such as which team should handle a request or whether a reply meets your checklist. Let it act alone only when it is very confident, and send close calls to a person. Microsoft says it must not be the only judge of decisions about people.