Decision 2.0
vLLM Semantic Router's Apache-2.0 decision models (0.6B to 27B) answering up to 64 typed questions per input in one forward pass.
This summary is based on how curated lists describe the project; its README was not available to review.
Overview
Decision 2.0 is a Hugging Face collection of open decision models from the vLLM Semantic Router project. According to the list summary, the models range from 0.6B to 27B parameters, are released under Apache-2.0, can answer up to 64 Choice, Noul and Score questions about one input in a single forward pass, and load with Transformers.
They are an open alternative to hosted Jev for teams that want to run typed decisions locally or inside a routing layer. No model card was reviewed here, so accuracy, calibration and hardware requirements are not stated.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Up to 64 typed Choice, Noul or Score questions about one input | — |
How it works
- Load a model size with Transformers
- Pass one input with many typed questions
- Read per-question probabilities from one forward pass
Strengths
- Open weights under Apache-2.0
- Many questions per forward pass
- Range of sizes
Caveats
- List-only entry; model card not reviewed
- Accuracy not stated here
Getting started
See the Hugging Face collection vllm-sr/decision-20.At a glance
- Type
- Model
- Status
- beta
- Language
- Python
- Runtime
- Transformers
- Has code
- yes
- License
- Apache-2.0
Use cases and integrations
How curated lists describe it
vLLM Semantic Router's Apache-2.0 decision models from 0.6B to 27B: up to 64 Choice, Noul, and Score questions about one input in a single forward pass, loadable with Transformers
AnotiaWang/awesome-decision-models
Frequently asked questions
Who publishes Decision 2.0?
The vLLM Semantic Router project.
How many questions per call?
Up to 64 Choice, Noul and Score questions in one forward pass, per the summary.
Do I need TypeSafe?
No, these are open weights loaded with Transformers.
Related projects
0.8B open System One model with task adapters that makes millisecond decisions in front of a larger local model like Qwen3.8-27B.
ModelbetaPythonClojure System One engine on Jolt that routes typed questions to Laya encoders, a llama.cpp 'thinker', Clef or TinyDecide.
ModelresearchClojureMangaba AI's local, calibrated typed-decision model for Portuguese text, trained to optimize the Brier score; proprietary since v5.
ProjectproductionPython- blink★ 22
Open one-pass typed-decision model with a dependency-free C99 runtime that also builds to a 66 KB WebAssembly module.
ServicebetaC - Foq★ 11
Local open alternative to Jev using a 1.58-bit ternary 8B model to answer booleans, choices, scores and Pydantic extraction in ~25 ms.
ProjectbetaPython Runs the open Laya decision model entirely in a browser tab with WebGPU and ONNX Runtime Web: no server, no API key.
ServicebetaTypeScript