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, as decision models become increasingly common on the web. The model, developed by AWS distinguished engineer Marc Brooker, is a small, fast, and low-cost model designed to make binary decisions and return confidence scores rather than generate text. Built on the "torso" of an LLM (specifically Qwen3.5-2B), it can run locally and is suited for agentic workflow steps where a fully-featured LLM is unnecessarily expensive or slow. Brooker noted that customers' workflows often didn't require the full capability of frontier LLMs for every step. The project originated as a homebrew effort that briefly topped the Jevbench rankings before being polished and released through Amazon's Strands Labs. The proliferation of similar models raises questions about their value, with Brooker highlighting the challenge of balancing decision accuracy with general-purpose intelligence. TypeSafe CEO Diogo Almeida downplayed competition, saying the current wave reflects machine learning enthusiasts implementing architectures rather than teams deeply focused on making intelligence useful, and noted that the difficulty of creating truly smart models may be underestimated.

(Source：TechCrunch)