Amazon Launches Open-Source Strands Decider 2B

Amazon Web Services put out its own version of a Jev model this week, joining a growing line of companies building small, narrow decision engines instead of chasing bigger general-purpose chatbots. The new release, called Strands Decider 2B, is open-source, small enough to run on a laptop, and built for one job: picking between a fixed set of options and reporting how sure it is about the pick. That's a different target than a frontier language model, which can write essays or debug code but costs more and runs slower for a task that's really just sorting.

The timing isn't an accident. OpenAI announced a similar offering in the same week, a sign that the appetite for cheap, fast decision-making tools inside AI agent pipelines has become real demand rather than a niche experiment. Amazon's version started as a side project from distinguished engineer Marc Brooker, who built it after seeing TypeSafe's original Jev model and wanting to try his own version. It did well enough on its own, briefly topping the Jevbench leaderboard for models its size, that Amazon's Strands Labs team picked it up and turned it into a product.