OmniJev/OneJev-27B

Hugging Face
VISIONPricing:Input $1.6 / Cached $0.15 / Output $12Concurrent Unit Cost:2Model Size:27BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 27, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Warm

OmniJev/OneJev-27B is a 27 billion parameter multimodal System One decision model developed by OmniJev. It is a full fine-tune of Qwen/Qwen3.8-27B, designed to process screenshots, photos, videos, and text inputs. The model returns calibrated probabilities for multiple options in a single forward pass, making it suitable for agentic decision-making tasks requiring rapid, probabilistic outputs from diverse inputs.

Loading preview...

OneJev-27B: A Multimodal System One Decision Model

OneJev-27B is a 27 billion parameter multimodal model developed by OmniJev, designed for System One decision-making. It is a full fine-tune of Qwen/Qwen3.8-27B, trained on 99,193 questions derived from real agent runs, videos, and images. This model excels at processing diverse inputs including screenshots, photos, videos, and plain text, providing calibrated probabilities for various options in a single forward pass.

Key Capabilities

  • Multimodal Input Processing: Handles images (screenshots, photos), video, and text simultaneously.
  • Probabilistic Decision Making: Outputs calibrated probabilities for multiple choices, ideal for agentic workflows.
  • High Speed Inference: Achieves rapid response times, answering 1 question in 189 ms and 10 questions in 324 ms (32.4 ms/question) on an H200 GPU with a 1280x720 screenshot.
  • System One API Compatibility: The server speaks TypeSafe's System One API, extended with a media field for rich inputs.
  • Flexible Deployment: Can be run with PyTorch via qev or on CPU/GPU using llama.cpp via GGUF builds (images supported, video requires PyTorch server).

Good For

  • Automated Agents: Powering agents that need to interpret visual and textual cues to make rapid, probabilistic decisions.
  • Task Automation: Automating workflows that involve analyzing screen content, images, or video streams.
  • Interactive Systems: Developing systems that require quick, context-aware responses based on multimodal input.
  • Probabilistic Reasoning: Applications where a calibrated probability for each action or outcome is crucial.

Finetunes
3 models