SGLang
SGLang's native /v1/decisions and /v1/systemone endpoints that read option probabilities from a validated chat model's next-token scores.
This summary is based on how curated lists describe the project; its README was not available to review.
Overview
This entry is a page in SGLang's documentation on decision models. According to the list summary, SGLang, the open-source LLM serving engine, provides native /v1/decisions and /v1/systemone endpoints that read option probabilities from a validated chat model's next-token scores, without needing a special decision checkpoint.
That makes SGLang a serving option for Jev-compatible clients against self-hosted open models. No page content was fetched, so the list of validated models, calibration behaviour and configuration details are not available here.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Other | Choice/Score/Noul questions served over /v1/systemone from next-token scores(inferred) | — |
How it works
- Serve a validated chat model with SGLang
- POST typed questions to /v1/decisions or /v1/systemone
- Read option probabilities derived from next-token scores
Strengths
- Uses a mainstream serving engine
- No special checkpoint required
Caveats
- Grounded only in a one-line summary
- Probabilities from chat models may need calibration
At a glance
- Type
- Service
- Language
- Python
- Runtime
- SGLang server
- Has code
- yes
Use cases and integrations
How curated lists describe it
Native `/v1/decisions` and `/v1/systemone` endpoints that read option probabilities from a validated chat model's next-token scores, with no special checkpoint
AnotiaWang/awesome-decision-models
Frequently asked questions
What does SGLang add?
Native decision endpoints compatible with the System One request shape.
Does it need a special model?
No; it reads probabilities from a validated chat model.
Where are the docs?
At docs.sglang.io under supported models, decision models.
Related projects
Native Go runtime serving Gemma 4 12B as a Jev-like decision model with hand-tuned PTX kernels and no CGo or CUDA toolkit.
ProjectbetaGoGo gateway exposing the Jev /v1/systemone contract over ordinary LLM backends by reading single-token label logprobs.
ServicebetaGoTurns any chat model on SGLang, vLLM, transformers or MLX into a Jev-compatible service with one prefill and one label token per question.
ProjectbetaPythonSelf-hosted Jev-compatible System One server that sits in front of llama-server and reads option probabilities from any GGUF model.
ServicebetaPythonInference engine for Jev-style decision models using segmented forwards with prefix reuse, serving the Jev wire contract.
ProjectresearchPythonResearch preview: Python client and pinned adapters for running System One decision models on your own Modal GPUs.
LibraryresearchPython