Benchmarks & research
60dB Judge is a commercial SystemOne-compatible decision model for AI agents, priced at $0.010 per million input tokens with free outputs.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
The vendor’s same-day test on the public JevBench set reports 214 correct decisions out of 231 for Judge and 199 for Jev 1.13.0. It is not an official submission to JevBench’s sealed set. The reported timings also leave out the internet round trip.
Judge runs on 60db’s hosted service. Give it a situation and ask it to choose, rate or answer yes/no from the answers you allow; it returns scores without writing a reply. You are using Judge, not TypeSafe’s Jev, and the vendor’s public-test results do not promise matching answers.