Amazon Releases Open-Source Decision Model Strands Decider 2B to Compete in Growing AI Niche
AWS’s Strands Decider 2B returns choices among predefined options with confidence scores, rather than generating free-form text, and runs locally at no cost. Built on Qen3.5-2B, it grew from Marc Brooker’s Jevbench-topping prototype.
Amazon Web Services released an open-source decision model called Strands Decider 2B in October 2026, entering a rapidly crowding field of so-called “Jev-like” models first popularized by TypeSafe’s Jev.
Unlike large language models that generate free-form text, Strands Decider 2B sorts between pre-defined options and returns a confidence score for its choice. Built on the Qen3.5-2B language model as its base, it is designed to run locally and is available now at no cost. The release came the same week OpenAI announced a similar offering.
The project originated with AWS distinguished engineer Marc Brooker, who built a personal version of the model after encountering TypeSafe’s Jev. That homebrew effort briefly reached the top spot on the Jevbench ranking for models of its size, prompting Amazon’s Strands Labs — a team focused on tools and protocols for AI agents — to formalize and release it publicly.
Brooker said the motivation came from conversations with AWS customers whose automated workflows didn’t always need a full-featured LLM. “What originally piqued my interest in this class of models was that they make a perfect decider for a workflow step,” he told TechCrunch, adding that the approach offers “lower latency, potentially lower cost” compared to standard models.
He cautioned, however, that optimizing speed without eroding the model’s broader knowledge remains a key challenge. “There is a very careful balance to be found,” Brooker said.
TypeSafe CEO Diogo Almeida, whose company coined the Jev concept — named after economist William Stanley Jevons — was measured about the competitive threat. “The current batch seems more like ML people wanting to implement a cool architecture than a team deeply dedicated to making intelligence useful,” he told TechCrunch.
The release of dozens of similar models since TypeSafe debuted the concept suggests broad interest in lightweight decision-making AI, though it also raises questions about how much differentiation is possible in the space.