Amazon Web Services has introduced an open-source decision model inspired by TypeSafe’s Jev, responding to the growing demand among AI developers for intelligence that is better suited to computer automation than large language models.
AWS's Strands Decider 2B, released just days before OpenAI announced a comparable offering, offers a high-speed, low-cost method for sorting through pre-determined options and providing confidence metrics for those choices. The model is fully open-sourced, publicly available, and compact enough to operate on local hardware.
Marc Brooker, an Amazon distinguished engineer, initiated the project after observing Jev and attempting to construct his own version of such a model. His internal prototype was successful enough to briefly rank at the top of the Jevbench for its size class, leading AWS engineers to refine it and release it as a product from Strands Labs.
According to Brooker, the necessity for this tool became apparent during discussions with AWS clients, whose agentic workflows did not always require the expense and capability of a full-scale LLM.
"What originally piqued my interest in this class of models was that they make a perfect decider for a workflow step— 'what is the next thing for me to do here, based on where I am?'" Brooker explained to TechCrunch. He emphasized that these models allow for a workflow step that is "more reliable," thanks to confidence scores and a restricted domain of answers, while offering "lower latency, potentially lower cost."
Similar to other decision models, Strands Decider utilizes the core architecture of an LLM—in this case, Qwen3.5-2B—but diverges by outputting calibrated choices rather than generating text. TypeSafe named their original model Jev after the economist William Stanley Jevons, referencing his theory regarding how falling costs can increase demand.
The proliferation of dozens of analogous models produced by researchers indicates significant industry interest, though the long-term value of this niche remains a subject of debate. Brooker suggested that the primary challenge lies in optimizing the model's speed without sacrificing its intelligence.
"There is a very careful balance to be found where you want to push its performance on accuracy and calibration on these kinds of tasks, without degrading its performance on understanding different languages, on having the kind of knowledge it has, which is what makes it general purpose and interesting and useful," he told TechCrunch.
Despite the trend, Brooker does not anticipate that major frontier labs will dominate this specific space, particularly since building a functional model can be achieved for a few hundred or thousand dollars.
TypeSafe executives stated that they are focusing on improving future iterations rather than engaging in a race. "I get that people think it’s a gold rush, but they might be underestimating the difficulty of making the models actually smart," CEO and founder Diogo Almeida told TechCrunch. He noted that for the moment, he does not perceive serious competition from other entities.
"The current batch seems more like ML people wanting to implement a cool architecture than a team deeply dedicated to making intelligence useful."
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.