Jev directory
CommunityJev Models & Reimplementations
Darkstar JEV 27B Decision ModelOpt W4A4 NVFP4 Mixed FP8
Hugging Face model repository hosting a 27B JEV decision model checkpoint quantized with ModelOpt in a W4A4 NVFP4 mixed-FP8 scheme. The repo carries the tokenizer and chat template needed to load and prompt the model.
- Reading time
- 2 min
- Published by
- Community
- Added
- Oct 2026
Highlights
- Repository name identifies a 27B JEV decision model quantized to W4A4 NVFP4 with mixed FP8.
- Ships a chat template that renders tool calls as XML <tool_call> blocks.
- Chat template exposes reasoning effort settings: xhigh (default), medium, and low.
- Template handles image and video content blocks alongside text.
- Hosted on Hugging Face under the HangGlidersRule account.
Questions about this entry
Short answers drawn from the linked project or page.
It is a Hugging Face model repository for a 27B JEV decision model checkpoint quantized with ModelOpt using a W4A4 NVFP4 mixed-FP8 scheme, published under the HangGlidersRule account.
Download the repository files from Hugging Face and load them with a runtime that supports the NVFP4 and mixed FP8 quantization formats. The repository includes the tokenizer and chat template needed for prompting.
Yes. The template renders available functions inside a <tools> block and expects calls in <tool_call> XML with a nested <function=...> block and <parameter=...> entries, and it rejects calls with a suffix after the closing tag.
The template accepts reasoning_effort values of xhigh (the default), medium, and low, and raises an exception for any other value. Each setting injects a different instruction into the system message.
The template includes branches for image and video content items, emitting vision start, pad, and end tokens. System messages that contain images or videos raise an exception instead of rendering.
The provided metadata does not state a license. Check the repository page on Hugging Face for the license and any usage restrictions before redistributing or deploying the weights.
If your runtime or hardware lacks support for W4A4 NVFP4 and mixed FP8 kernels, or if you need full-precision weights for further training, use an unquantized JEV checkpoint instead.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Standard One SH (StandardThinking)
LinkedIn: What multimodal decision models actually change, and what they do not.
LinkedInJevDirectory.org
Hacker News: Decision Index 0.2.1 scores open reproductions of Jev on a frozen suite.
jackbravo4 pointsDecision Index 0.2.1 – 70 open reproductions of TypeSafe Jev's Decision ModelRead the thread on Hacker NewsHacker News: Pointer to a DiffusionGemma plus vLLM tutorial for a Jev-shaped classifier.
1 pointHow to build a Jev-style classifier with DiffusionGemma and vLLMRead the thread on Hacker NewsMore Jev resources like this
A native MLX runtime for Laya's open typed-decision weights: short decisions on Apple Silicon, with no generated tokens, PyTorch, or cloud API.
Jev Models & Reimplementations#community#python#open-models
Community
mizorewww/laya-mlxPython
mizorewww/laya-mlxPythonAn open-model semantic-if engine positioned as a local Jev-style decision stack, separate from the original SemIf repo but sharing the same author.
Jev Models & Reimplementations#community#python#open-models
Community
TheoLeeCJ/SemIf-OpenJevPython
TheoLeeCJ/SemIf-OpenJevPythonLaya typed decisions through Apple Core ML and the Neural Engine, with the author's speed and energy benchmarks for short decisions on an M3 Max.
Jev Models & Reimplementations#community#python#open-models
Communitymizorewww/laya-coremlPython
mizorewww/laya-coremlPythonA local yes/no decision model that speaks Jev's wire format on a laptop CPU, refusing Choice and Score until those question types ship.
Jev Models & Reimplementations#community#python#open-models
Community
feder-cr/jevC++
feder-cr/jevC++A local Jev-compatible POST /v1/systemone server in TypeScript that maps typed questions onto oMLX classification prompts and normalizes logprob-style answers.
Jev Models & Reimplementations#community#typescript#local
Community
githubnext/localjevTypeScript
githubnext/localjevTypeScriptA local daemon that pulls open decision models and serves them on TypeSafe's /v1/systemone, so a Jev client switches hosts with TYPESAFE_BASE_URL.
Jev Models & Reimplementations#community#rust#open-models
Community
ollaya-dev/ollayaRust
ollaya-dev/ollayaRustFrom the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.