1.4k likesNew · 24d old
AutoTrust-JEV-27B scores 67 on Decision Index, excels at tool selection
What is AutoTrust-JEV-27B-v1.1?
AutoTrust-JEV-27B is a 27-billion-parameter model that evaluates multiple-choice questions by returning probability scores for each answer option. It’s tuned for structured decisions like intent classification, tool picking, and simple automation tasks.
Who it is for
Developers building assistants or workflow bots can use it to rank actions or responses with calibrated confidence.
The story so far
Early Signal first spotted autotrust/AutoTrust-JEV-27B-v1.1 on November 8, 2026, with 1.4k likes. On its busiest day it gained 60 likes. The project is 24 days old. Early Signal files it under MCP & Skills.