NVIDIA Open-Sources PixelUMM, a 15B Encoder-Free Unified Multimodal Model

NVIDIA researchers have open-sourced PixelUMM, a 15B-parameter encoder-free unified multimodal model that handles both image and video understanding and generation directly in raw pixel space, eliminating the need for a VAE or pretrained vision encoder.

2026-10-05 ~ 2026-10-06 · 2 related posts

Full story(2 episodes)→

• Episode 1: NVIDIA Releases PixelUMM, Encoder-Free Unified Multimodal Model(2026-10-02, 3 posts)
• Episode 2: NVIDIA Open-Sources PixelUMM, a 15B Encoder-Free Unified Multimodal Model(2026-10-05, 2 posts)

• NVIDIA Open-Sources PixelUMM: Encoder-Free Model Handling Image/Video in Raw Pixel Space — udmrzn · 2026-10-05
• NVIDIA's 15B PixelUMM drops VAE and vision encoder, learns image understanding and generation from raw pixels — lmoroney · 2026-10-06