PixelUMM
PixelUMM is an encoder-free multimodal model for understanding and generation across text, images, and video directly in pixel space. It accepts text, RGB images, and video frames, and produces text, RGB images, and video frames.
What it does
Researchers and developers can use PixelUMM to study unified multimodal modeling, generate images or video from text prompts, and produce text conditioned on images or video.
- Availability
- Released checkpoint on Hugging Face
- Pricing model
- Not established
- Deployment
- Requires Linux, an NVIDIA GPU, CUDA build of PyTorch, and FlashAttention
- Privacy
- Not established
- Licence
- NVIDIA One-Way Noncommercial License; use limited to non-commercial research or evaluation
Capability
Modality
Access
Deployment
Pricing
Target user
Evidence
6 claims- PixelUMM supports text-to-image generation and image-conditioned text generation. [1]
- PixelUMM supports video-conditioned understanding and video generation. [1]
**PixelUMM: Encoder-Free Unified Image and Video Understanding and Generation**
NVIDIA. Quoted text is outside CC BY 4.0 and keeps its original rights. - The PixelUMM model checkpoint is released on Hugging Face.
- PixelUMM requires Linux, an NVIDIA GPU, a CUDA build of PyTorch, and FlashAttention. [1]
PixelUMM requires Linux, an NVIDIA GPU, a CUDA build of PyTorch, and
NVIDIA. Quoted text is outside CC BY 4.0 and keeps its original rights. - The checkpoint is licensed for non-commercial research or evaluation purposes only. [1]
> PixelUMM is released for non-commercial research or evaluation purposes only.
NVIDIA. Quoted text is outside CC BY 4.0 and keeps its original rights. - PixelUMM has not been validated for production or high-stakes use. [1]
PixelUMM has not been validated for production or high-stakes use.
NVIDIA. Quoted text is outside CC BY 4.0 and keeps its original rights.
Sources
1
1
Verified
Versions
3| Version | Effective | Change | Changed fields | State |
|---|---|---|---|---|
| 3 | 2026-10-15 20:00 UTC | material change | availability | Current |
| 2 | 2026-10-03 07:47 UTC | material change | availability | Current |
| 1 | 2026-10-02 08:40 UTC | publication | None | Current |
In briefs
2- Backburner leads 25 AI developments in week 40, 2026 Weekly brief 2026-W40
- Google introduces Gemini 4 Argon reasoning model Daily brief 2026-10-02
Original catalog content is licensed under CC BY 4.0, attributed to wowsuchai. Source quotations and linked content keep their original rights.