JevBench: Microsoft-Decision-1 ranks #6 among API decision models, #12 overall
Florian S @airesearch12X
Why it matters
Independent (community) benchmark result that contrasts with Microsoft’s own ‘top performance’ claim.
Summary
The author of the community benchmark JevBench (benchmarkheaven.com) reports that Microsoft-Decision-1 ‘scores #6 in its peer group of API provided decision models’ and ‘#12 on the composite score’ in a mixed leaderboard of open and API models. Jev is #4 and Sage #1 among API models. It is strong in recruiting, risk assessment, gaming, advertising, science and moderation, and ‘the strongest API provided model in the top 10 in mixed-language tasks’. Unofficial benchmark.
Metrics read 2026-10-12 via api.fxtwitter.com: 15,992 views, 167 likes.
Archived text
Official JevBench results:
Microsoft-Decision-1 scores #6 in its peer group of API provided decision models.
Compared with Jev it performs strong on many typical use cases, most notably on Recruiting, Risk assessment, Gaming, Advertising, Science and Moderation.
It’s also the strongest API provided model in the top 10 in mixed-language tasks.
In our headline ranking for API offered models Jev superseeds it on #4 and Sage leads on #1.
Please be aware that JevBench seperates the leaderboards of
- API offerered models such as Microsoft-Decision-1 (found here: https://benchmarkheaven.com/jev-models/api) and
- open weights models (found here: https://benchmarkheaven.com/jev-models), because speed and cost can only be scored in a fair way within these seperate peer groups.
In a mixed leaderboard of open and API hosted models it lands #12 on the composite score.
It’s a good model, and certainly will find its customer base, given Microsofts strong position in enterprise compliant cloud offerings.
Congrats @Microsoft!
Quoting @satyanadella: Introducing Microsoft-Decision-1, our new model for fast decision-making.
It delivers top performance on structured decision tasks, outperforming both LLMs and other decision models in latency and quality.
We’re already testing it across Microsoft for everything from incident response and quality control to scientific discovery.
