AWS Strands Labs Releases Strands Decider 2B: An Open Source Decision Model That Picks Options in About 115 ms
115 ms. AWS Strands Labs just dropped a 2B decision model that routes agent choices faster than most retrieval calls.
Why it matters
AWS released Strands Decider 2B, an open-source decision model built on Qwen 3.5-2B that returns structured choices with calibrated confidence in a single forward pass—no text generation. At 115 ms median latency on consumer GPU and 0.723 JevBench score, it's positioned as a fast local option for agent routing, tool selection, and guardrails. The model's architecture (structured output, no text, calibrated confidence) and public benchmark make it a concrete alternative to text-generating models for agent control flow.
The key facts
6 to knowStrands Decider 2B built on Qwen 3.5-2B-Base, Apache 2.0 licensed
115 ms median latency on RTX 3090
0.723 score on JevBench public benchmark
Returns choices, yes/no probabilities, and scores in one forward pass; never generates text
Designed for routing, tool selection, and guardrails in agents
AWS Strands Agents team release
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: AWS's Strands Agents team released Strands Decider 2B, an Apache-2.0 decision model built on Qwen3.5-2B-Base. It returns choices, yes/no probabilities and scores with calibrated confidence in one forward pass, never text. It runs at a 115 ms median on an RTX 3090 and scores 0.723 on the JevBench…