Strands Decider 2B: a small, open-source, decision model

Strands Decider 2B is a small open model that picks among options rather than writing text. Given a question and a list of options, it scores each option, and it can also give a reliability score for each decision, for example on yes or no questions. According to the Strands announcement, it starts from Qwen3.5-2B, a 2-billion-parameter model from Qwen, removes the text-generation output, adds a small scoring head, and is fine-tuned with a LoRA adapter, a lightweight add-on to a model's weights.

The developers report median latency around 115 ms on a local RTX 3090 and about 153 ms on an M3 MacBook for small tasks, and a rank of 3rd of 33 models in its 2-billion-parameter class on the public set of a benchmark called JevBench, measuring accuracy and calibration. They state its limits: it is significantly worse than reasoning models on complex problems and cannot generate text, so it is unsuited to coding or chat. All results are the developers' own.