Amazon Web Services has released Strands Decider 2B, an open source decision model inspired by TypeSafe’s Jev, as AI developers increasingly seek intelligence optimized for computer automation rather than general-purpose text generation.
What Happened
Strands Decider 2B is a high-speed, low-cost model designed to sort between pre-decided options and deliver a confidence score for its choice. The model is fully open source, available immediately, and small enough to run locally. It was developed by Amazon’s Strands Labs, an organization focused on new tools and protocols for deploying AI agents.
The project originated with Amazon distinguished engineer Marc Brooker, who built a homebrew version after encountering TypeSafe’s Jev. This initial prototype briefly reached the top spot on the Jevbench ranking for models of its size, prompting Amazon engineers to refine and release it. Strands Decider 2B is built on the 'torso' of the Qwen3.5-2B LLM but is trained to deliver calibrated choices instead of generating text.
The release coincides with a similar offering announced by OpenAI the same week. TypeSafe, the company behind the original Jev model, named their creation after economist William Stanley Jevons to invoke his theory that falling costs can increase demand. Since TypeSafe’s debut, dozens of similar models have been produced by researchers.
Why It Matters
The emergence of specialized decision models addresses a specific gap in agentic workflows identified by AWS customers. According to Brooker, many workflows do not require the full capability or cost of a frontier LLM for every step. He notes that decision models offer 'a workflow step that can be structured in a way that is more reliable, thanks to the confidence scores, thanks to the closed domain of answers, [and is] lower latency, potentially lower cost.'
Brooker emphasizes that the primary technical challenge lies in balancing performance. He states, '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.'
Despite the influx of competitors, Brooker does not expect frontier labs to dominate this niche, noting that the cost to build interesting models in this smaller market is in the hundreds or thousands of dollars. Conversely, TypeSafe CEO Diogo Almeida suggests that current competitors may be underestimating the difficulty of making these models truly smart, stating, 'The current batch seems more like ML people wanting to implement a cool architecture than a team deeply dedicated to making intelligence useful.'
The Bottom Line
Amazon’s release of Strands Decider 2B signals growing industry interest in lightweight, specialized models for agentic automation. While the space is rapidly filling with clones and variants inspired by TypeSafe’s Jev, developers face a trade-off between raw intelligence and the reliability, latency, and cost efficiencies required for specific workflow steps.