Amazon Web Services Inc.'s Strands Labs team has created an open-source decision model called Strands Decider 2B. The company made the project available to the open-source community to accelerate agentic AI development. Decision models make rapid choices without generating text.
Traditional large language models respond to inputs by generating text, code, images, or video. Decision models make choices when presented with predefined options and assign a confidence score to each response. Skipping text generation makes these systems fast and reduces latency.
Decision models gained attention after startup TypeSafe AI Inc. released Jev. AWS believes Jev has structural shortcomings, including performance issues during intricate reasoning. Strands Decider 2B is AWS's attempt to improve on that concept.
Read nextGoogle Releases Gemini 4 Argon Frontier Model to Close Gap With RivalsThe model runs with low latency
The new model uses the Qwen3.5-2B base torso with a customized pointer head containing one million parameters. AWS fine-tuned the model using a rank-16 LoRA adapter. The two billion parameter scale allows the model to run locally on a machine with under 150 milliseconds of latency.
Developers can download Strands Decider 2B via Hugging Face. The codebase, training scripts, and examples are available on GitHub.
