AI Briefing
KOSign in

Vimoo 1.0 Open-Source Video Generation Model Family Released

·2026.11.15 23:24

Key point

The family includes a 9.5B Lite variant and supports ComfyUI, Diffusers, and MLX runtimes.

Details

  • Vimoo 1.0 is an open-source video generation model family supporting text-to-video (T2V) and image-to-video (I2V) tasks.
  • The main Vimoo 1.0 model uses approximately 11.7B parameters per task, while the Vimoo 1.0 - Lite variant uses approximately 9.5B parameters.
  • Available on Hugging Face for Diffusers, ComfyUI, and MLX (Apple Silicon) runtimes, with distilled versions available for ComfyUI and MLX.
  • The Lite variant supports 720p resolution at 24 fps using 32x32x4 spatiotemporal compression.
  • Features include a dedicated text encoder for prompt adherence, an optional prompt enhancer, and training on curated aesthetic data for cinematic quality.
  • Distilled LoRA adapters for T2V, I2V, and TI2V are not interchangeable between tasks.
open-sourcevideo-generationmlxtext-to-videocomfyuiimage-to-videovimoo
Read originalShare on Threads

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.

Related

RedditMeituan releases LongCat-Video-Avatar 1.6 for stable long-form audio-driven video generationRedditVellium 1.9.5 adds Lite mode to run visual novels on 12B local modelsGeekNewsraea: Open-source engine renders AI-generated React/TypeScript code to video, 3D, and games on local GPUPyTorchKRllama.cpp v1.0.0, First Major Release in Three and a Half YearsGeekNewsGitHub Next releases Ace, a local collaborative coding environment for humans and AI agentsCloudflare BlogCloudflare open-sources Kitesurf, its stateless browser for AI agents