Baseten releases joey-1 for low-cost typed decisions
Base Labs released joey-1, a two-billion-parameter decision model post-trained from Qwen3.5-2B. It accepts text or structured state plus typed yes/no, multiple-choice or ordinal questions and returns calibrated probabilities without free-form generation; weights and a hosted Baseten Model API endpoint are available.
Why it matters: Routing, moderation and approval workflows often need a bounded decision rather than prose. A small model with typed outputs can cut latency, output parsing and token cost, but calibration, bias, domain transfer and override design matter more than vendor…