Strands Decider: AWS's open-weight decision model that picks instead of generating
Open-weight decision models from AWS Strands Labs that choose from a fixed set of options and return calibrated confidence scores instead of generating text, built for routing, triage, and guardrail checks inside agents.
Why it matters
Amazon's internal teams replaced Claude Sonnet 4.6 with the 4B Gemma-based Strands Decider E4B for ticket labeling, matching production accuracy at 98% lower cost and 78% faster calls: a real production proof that tiny decision models can displace frontier LLMs on narrow judgments.
What you could build with it
A support team drowning in ticket triage costs could route every incoming ticket through a small self-hosted decision model that classifies intent and confidence, escalating only low-confidence cases to a frontier model and cutting the AI bill by an order of magnitude.
Does it hold up?
Usable now for narrow decisions: open weights on Hugging Face under Apache-2.0, with a production deployment inside Amazon cutting ticket-labeling cost 98%; the package is still an experimental 0.x release and Gemma 4 support is on main, not yet in the PyPI build.
Built with Strands Decider
- Amazon saves millions by replacing Sonnet 4.6 with Strands Decideragxsearch · Amazon engineers replaced Sonnet 4.6 with Strands Decider E4B for internal ticket labeling, performing just as well at 98% less cost and saving millions in token costs per year.
- Connecting decision models to Amazon Bedrock AgentCore Evaluationsmechnext · How to wire decision models into Bedrock AgentCore Evaluations as code-based evaluators, using Strands Decider 2B as the self-hosted backend.
- Amazon rewrites its software engineering playbook for an era when AI writes the codegeekwire · Velocity Labs, the AWS team behind Strands, built Strands Decider 2B in under a week from concept to public release.
Learn more
First spotted on article: source.
More AI models
VoiceStudio: open-source, fully-local ElevenLabs alternative for voice cloning, dubbing, and audiobooksdebpalash/VoiceStudio is a desktop app that runs voice cloning, voice design, video dubbing, dictation, transcription…models · JEV 0.75Edge0-145B: open 145B MoE that runs on a single consumer GPUEdge0 open-sourced 145B and 120B latent-MoE models derived from Qwen3.8-Flash-Next, plus an inference framework that…models · JEV 0.74Strata: run a 125B model on a gaming PCFree, MIT-licensed local inference engine that runs Qwen3.8-Flash-Next (125B MoE, about 6B active per token) on…models · JEV 0.73Google AI Edge Gallery: offline speech recognition and synthesis on-deviceGoogle added fully offline speech recognition and text-to-speech to AI Edge Gallery, with ML Kit APIs on Android and…models · JEV 0.73Soup: fine-tune an 8B LLM on a 4 GB laptop GPU from one YAML fileAn open-source CLI that fine-tunes LLMs from a single YAML file, with layer streaming that trains an 8B model on a 4…models · JEV 0.73Tencent Hunyuan Hy3: open 295B MoE with 21B active under Apache 2.0Tencent open-sourced Hy3, a 295B-parameter mixture-of-experts model with only 21B parameters active per token, under…models · JEV 0.73
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.