Amazon unveils a free, fast, open source Jev killer: Strands Decider 2B makes decisions in fractions of a second

The AI world has spent two years obsessed with models that write. Now the hottest new category is models that *choose*. Amazon’s Strands Decider 2B, covered by VentureBeat, is part of a fast-growing wave of “decision models” that skip the whole essay-writing step and go straight to picking an option. That’s a massive shift in how agents will actually work in production: less chatbot, more control plane. This is ultimately an economics story. Every startup building AI agents knows that cost and latency are the two killers. A model that makes decisions in fractions of a second — and is open source and free to download — undercuts the old “generate a paragraph, then parse it” approach. Amazon entering this space signals that even the biggest cloud players see decision-making as a strategic layer worth owning, not just another inference workload. The speed of this wave is striking. TypeSafe AI introduced Jev just two weeks ago, and already we’re seeing a rush of competitors, many with open weights. That’s the Llama/Mistral playbook applied to agent reasoning. The likely outcome? Decision models become commoditized fast, and the real moats shift to orchestration, data pipelines, and evaluation frameworks. As VentureBeat’s coverage highlights, the race is moving beyond model quality to raw operational efficiency. Still, we should keep some healthy skepticism. “Fast and free” is great, but accuracy on real-world agent tasks is the true test. If Strands Decider 2B can hold up on messy, ambiguous decisions — not just clean multiple-choice scenarios — then Amazon may have just reset the baseline for everyone else. Either way, the decision-model era is here, and it’s moving at startup speed. Source: [VentureBeat](https://venturebeat.com/technology/amazon-unveils-a-free-fast-open-source-jev-killer-strands-decider-2b-makes-decisions-in-fractions-of-a-second)