Amazon releases its own Jev clone as decision models flood the web

Amazon Web Services has released Strands Decider 2B, an open-source decision model inspired by TypeSafe's Jev, through its Strands Labs organization. The release came the same week OpenAI announced a similar offering.

According to the source, Strands Decider 2B is a high-speed, low-cost way to sort between pre-decided options and deliver a measure of how confident it is in its choice. It is fully open-sourced, 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 such a model. The source states the homebrew project briefly reached the top spot on the Jevbench ranking for models of its size, after which Amazon engineers cleaned it up and released it as an offering from Strands Labs, described as an organization developing new tools and protocols for deploying AI agents.

Brooker told TechCrunch the need emerged in conversations with AWS customers whose agentic workflows did not always require the capability or cost of a fully-featured LLM. He said the model offers customers a workflow step that can be structured more reliably, thanks to confidence scores and a closed domain of answers, with lower latency and potentially lower cost.

Like other decision models, Strands Decider is built on the torso of an LLM, in this case Qen3.5-2B, but delivers calibrated choices instead of generating text. TypeSafe named its model Jev after economist William Stanley Jevons. Brooker said the challenge will be optimizing speedy decision-making without compromising intelligence, balancing accuracy and calibration against language understanding and knowledge.

He said he does not necessarily expect frontier labs to dominate the space, since smaller markets mean the cost to build something interesting is in the hundreds or thousands of dollars. TypeSafe CEO and founder Diogo Almeida told TechCrunch he did not see real competition emerging yet and said the current batch seems more like ML people implementing a cool architecture than a team deeply dedicated to making intelligence useful.