AgentPlays
AutoTrust releases JEV-35B, an open-weight decision model with about 3B active parameters
autotrust/JEV-35B is an Apache-2.0 open-weight decision model fine-tuned from Qwen3.5-35B-A3B, a 35B mixture-of-experts with about 3B parameters active per token. It returns calibrated choices and scores for ranking and classification, picking among up to 256 options in one pass. A separate NVFP4 build quantizes the routed experts to cut weight memory to about 23 GiB.
How you’d use it: Use a small-active open model to rank or classify items (search results, tickets, feeds) on your own GPU.