Open-Source ComfyUI Nodes: Integrate Sora, Veo, and Other Closed Models into Local Workflows
Fun_Walk_4965 · reddit · 2026-08-10
Since top-tier video models like Sora 2, Veo 3.1, and Kling 3 cannot be run locally, developers often switch between ComfyUI and multiple browser tabs.
To solve this, the author open-sourced a ComfyUI node pack that calls these hosted closed-API models directly from within a ComfyUI graph. The plugin supports task groupings like text-to-image, image-to-video, multi-reference video, and editing. Its core advantages include: keeping local and cloud models chained in the same workflow, supporting parallel generation (e.g., 3x parallel text-to-video), and using multi-reference nodes to maintain character consistency across shots.
More from coding & agent
- Cutting Costs in Autonomous Coding: Use Expensive Models as Mentors, Not Supervisors — Dan_Jeffries1 · 2026-08-10
- Learn Rust with 100 Exercises: A 9.4k-Star GitHub Course — tom_doerr · 2026-08-10
- Deploying Itself: Muse Glimmer Agent Demo — ben_burtenshaw · 2026-08-10
- Testing Meta's Muse Glimmer: Getting the 30B Model to Quantize Itself — ben_burtenshaw · 2026-08-10
- Research Agent for Muse Glimmer Live on Hugging Face Hub — ben_burtenshaw · 2026-08-10
- Meta Releases Muse Glimmer: A 30B Open-Weight Model for Local Agents — AIatMeta · 2026-08-10