Amazon Enters the Decision Model Market with Strands Decider 2B

The project originated from distinguished engineer Marc Brooker, who developed a prototype after observing the success of TypeSafe’s Jev model. After his initial version climbed the Jevbench rankings, the team at Strands Labs refined the architecture for a broader release. Built on the Qen3.5-2B model, Strands Decider functions by selecting from a pre-defined set of options while providing a confidence score for each choice, a departure from the open-ended nature of traditional language models.

Brooker identifies the primary utility of these models in their ability to act as reliable, low-latency decision points within complex agentic workflows. While the rise of similar models signals a surge in developer interest, industry incumbents remain skeptical of the new entrants. TypeSafe CEO Diogo Almeida dismissed the wave of clones as a superficial trend, suggesting that many current projects prioritize architectural novelty over the difficult work of achieving genuine, robust intelligence. Despite the skepticism, Amazon’s move reflects a broader industry shift toward specialized, local-running tools that bypass the heavy costs associated with large-scale frontier AI.