Jev directory
CommunityJev Models & Reimplementations
Jev Vision 2B GGUF
Hugging Face model repository from nikolaus714 distributing a Jev vision model in GGUF format. Its chat template accepts image and video content blocks, wrapping them in vision start and end tokens for multimodal prompts.
- Reading time
- 2 min
- Published by
- Community
- Added
- Nov 2026
Highlights
- Hosted on Hugging Face as nikolaus714/Jev-Vision-2B-GGUF
- Distributed in GGUF format for local inference runtimes
- Chat template renders image content with vision start and end tokens
- Template also handles video content blocks and rejects unexpected content types
- Repository name indicates a 2B-parameter vision variant
Questions about this entry
Short answers drawn from the linked project or page.
It is a Hugging Face model repository published under the account nikolaus714. The listing distributes a Jev vision model in GGUF format, and the repository name indicates a 2B-parameter vision variant.
The repository is named with the GGUF suffix, indicating the weights are distributed in GGUF format, the container used by local inference runtimes that load quantized models.
The chat template in the listing renders image content blocks between vision start and vision end tokens, and separately handles video content blocks with a video pad token. System messages containing videos raise an exception in the template.
The listing material does not include run or installation instructions. GGUF files are typically loaded by local inference runtimes that support the GGUF container, so check the repository page for the current usage notes.
The provided listing material does not state a license for this repository. Check the model page on Hugging Face for the license field before redistributing or using the weights.
It is published under the nikolaus714 account on Hugging Face. The listing material does not name any other maintainer, organization, or upstream project behind the release.
The template raises exceptions for system messages that contain videos, for unexpected item types inside content, and for unexpected content types. It also requires at least one message and a user query in multi-step tool conversations.
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 NewsFrom 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.