Amazon Web Services (AWS) has released Strands Decider 2B, an open-source decision model. This new offering from Strands Labs is designed for computer automation, providing a high-speed, low-cost method to sort between pre-decided options and indicate its confidence in a choice. It is small enough to run locally and is fully open-sourced.
The project was initiated by Amazon distinguished engineer Marc Brooker, who developed his own version after encountering TypeSafe's Jev. This homebrew project achieved success, briefly topping the Jevbench ranking for models of its size, leading to its release by Amazon engineers.
Brooker stated that the need for such a tool arose from discussions with AWS customers, whose agentic workflows did not always require the full capabilities or cost of a large language model (LLM). Strands Decider 2B is built on the "torso" of an LLM, Qen3.5-2B, but focuses on delivering calibrated choices rather than generating text.
TypeSafe CEO and founder Diogo Almeida commented on the influx of similar models, suggesting that many might be underestimating the difficulty of making these models truly intelligent. He indicated that TypeSafe is concentrating on improving its future models.