Amazon releases an open source decision model to rival Jev

Amazon Web Services has released an open source decision model inspired by TypeSafe’s Jev, as AI developers increasingly look for intelligence suited to computer automation rather than frontier large language models. The new model, Strands Decider 2B, arrived the same week OpenAI announced a similar offering. It is a high-speed, low-cost way to sort between a set of pre-decided options and return a measure of how confident it is in its choice. The model is fully open source, available now, and small enough to run locally.

Amazon distinguished engineer Marc Brooker came up with the project after seeing Jev and trying to build his own take on the model type. The homebrew effort was successful enough — it briefly reached the top spot on the Jevbench ranking for models of its size — that Amazon engineers refined it and released it through Strands Labs, the company’s organization for building tools and protocols for deploying AI agents.

Brooker said the need for this kind of tool emerged from conversations with AWS customers whose agentic workflows didn’t always require the capability or cost of a fully featured LLM. He described the model as a reliable decider for a workflow step, offering lower latency and potentially lower cost thanks to its confidence scores and closed domain of answers. Like other decision models, Strands Decider runs on the “torso” of a larger model — in this case Qwen3.5-2B — but instead of generating text, it delivers calibrated choices. TypeSafe named its own model Jev after economist William Stanley Jevons, invoking his theory that falling costs for something, including computer intelligence, can increase demand for it.

The number of similar models produced by researchers since TypeSafe introduced the idea shows the wide interest in the category, but also raises the question of how valuable the models can ultimately be. Brooker said the challenge will be optimizing speed without degrading the model’s broader language understanding and knowledge. He does not necessarily expect frontier AI labs to dominate the space, noting that building something useful in this smaller category can cost just hundreds or thousands of dollars.

TypeSafe, for its part, says it is focused on improving its own models rather than reacting to competitors. “I get that people think it’s a gold rush, but they might be underestimating the difficulty of making the models actually smart,” TypeSafe CEO and founder Diogo Almeida told TechCrunch, adding that he does not yet see real competition for his company.

Why it matters

The rapid spread of smaller “decision models” suggests AI developers are moving past one-size-fits-all LLMs toward cheaper, purpose-built components for routine automation tasks — a shift that could reshape the cost structure of deploying AI agents at scale, even as it remains unclear how many players beyond TypeSafe and Amazon can make the category technically sound.