INT4 Convrot Comfy Models is a collection of INT4 ConvRot-quantized models designed for use with ComfyUI. The repository includes diffusion, video, and upscaling models that are tailored to operate efficiently on GPUs with approximately 8GB of memory, such as the RTX 3070 Ti, while maintaining output quality. The models are intended to facilitate image and video generation as well as upscaling tasks within the ComfyUI ecosystem.
1 for distilled video diffusion, Sulphur 2 Base for video diffusion, and SeedVR2 (7B) for both image and video upscaling. Each model is quantized using the INT4 ConvRot method, which is aimed at reducing memory requirements without significantly compromising performance. Some models are described as distilled or turbo variants, indicating optimizations such as fewer processing steps or builds based on previous checkpoints.
This suite of models is suitable for users of ComfyUI who need to perform generative tasks on hardware with limited GPU memory. The models are distributed via a repository format, allowing users to access and implement them as needed for their generative workflows.
INT4 Convrot Comfy Models sits in PulseGate's Other AI category. It focuses on enabling efficient image and video generation and upscaling on consumer GPUs with limited memory using quantized diffusion models. It is built as an open-source project for AI developers and researchers. INT4 Convrot Comfy Models is open source under the MIT license. INT4 Convrot Comfy Models is available on the web, API, and the command line, and it can be self-hosted.
milo01 builds and maintains INT4 Convrot Comfy Models, and the product first shipped in 2026. Development happens publicly on GitHub with 46 stars and 22 commits in the last 90 days. Key capabilities include image generation, video generation, and upscaling.
Latest indexed changes and source events
milo01/INT4-Convrot-Comfy-Models verified by the PulseGate indexer
Other apps tracked under the same category.
