Efficient Large Model (NVIDIA and MIT) Open Models: Model List, Licenses and Articles

September 27, 2026

About Efficient Large Model (NVIDIA and MIT)

According to its organization card, Efficient Large Model is a collaboration between researchers from NVIDIA and MIT developing GPU-accelerated, efficient large models for generative AI. It releases mainly image and video generation models on its Efficient-Large-Model organization on Hugging Face.

Aiming for models that are “not only powerful but also efficient", its research centers on producing high-resolution images and video with little compute.

What It Releases

• SANA: a series of efficiency-focused diffusion models for image generation.
• SANA-Video 2.0 (August 2026): according to its model card, an efficient diffusion transformer for high-resolution video generation; the 5B-class model makes about eight seconds of 720p video from text or from text and an image. There is also a fast 4-step version.
• LongLive 2.0 (May 2026): a 5B model for long video generation, with NVFP4 builds.
• World models and streaming: research models such as SANA-WM and SANA-Streaming that keep generating video continuously.
• Research models such as Fast-dDrive for autonomous driving and Fast_dVLM, a diffusion-based vision-language model, are also available.

Licenses

Most, including SANA-Video, are Apache-2.0. LongLive 2.0 uses NVIDIA’s open model license, and some research repositories have no license field. Read each model’s license text before use. This summary is not legal advice.

Running Them Locally

• There are no official GGUF builds (as of 2026-09-26). Most models are used through diffusers or dedicated inference code.
• The 5B-class video models are a size you can try on consumer GPUs. They also load a text encoder and VAE, so check the memory needed in the tables in our articles.

Sources: the Efficient-Large-Model organization card; the model card of Efficient-Large-Model/SANA-Video_2.0_5B_720p; the model list of the Efficient-Large-Model organization on Hugging Face (all as of 2026-09-26).

Our Coverage and Data

Local Model Watch has published 1 article(s) on Efficient Large Model (NVIDIA and MIT)’s own models, 0 on third-party fine-tunes and quantized builds based on them, and 0 from its official blog. The model list below is collected by code from Hugging Face every day. Part of our publisher index; model families link to their family pages.

At a Glance

Recent Models on Hugging Face

The publisher lists 140 models in total on Hugging Face. Below are up to 15 of the most recent generative models (text, image, video, audio and so on), the scope of this site, per organization. Classifiers, feature extractors, research components and repositories without a model card are omitted, and the publisher’s own quantized builds and helper variants are folded into the row of the original model. Parameter counts are left blank for repositories that pack weights into integer types, where Hugging Face’s count is not the real parameter count.

From the Hugging Face API. Last changed 2026-09-27 (JST). Collected by code.

By Model Family

Articles on Its Own Models

Last updated 2026-09-27 (JST). The explanation at the top of this page was written with the help of AI from the primary sources it cites. The tables and lists under “Our Coverage and Data" are assembled by code from our article log and the Hugging Face API.