Radar / AI media tools / HyperFlow H3 Flash
HyperFlow H3 Flash: MiniMax H3 video squeezed to 13.7B for a single consumer GPU
Video Rebirth compressed MiniMax's 31B H3 video model to 13.7B via block pruning, low-rank factorization and INT8 quantization, sampling video with synced stereo audio in 4–8 steps from a 9.1GB GGUF on a 24GB card.
Why it matters
The first H3 build ordinary creators can actually run: a 9.1GB GGUF that generates 5–15s clips with audio on one RTX card — though the headline 12.7x speedup did not survive independent measurement.
What you could build with it
A small video studio could run HyperFlow H3 Flash in ComfyUI on a single 24GB GPU to draft short product clips with synced audio locally, iterating on prompts without renting datacenter GPUs or paying per-second video API fees.
Does it hold up?
Mixed evidence, honestly: it does run on a 24GB GPU (92s for a 5.2s clip on an RTX 5090 per the authors), but respected ComfyUI benchmarker Kijai measured only ~10% faster than the pruned full model, LoRAs from the 31B model do not load, and the MiniMax community licence excludes the EU, UK, US and Korea without authorization.
Built with HyperFlow H3 Flash
- HyperFlow H3 Flash Promises a 12x Speedup for MiniMax H3 Video — the Community Measured 10%newsnix.com · Contrasts Video Rebirth's 12.7x claim with Kijai's independent ComfyUI benchmark of roughly 10% over the pruned full model.
- HyperFlow H3 Flash ComfyUI: GGUF Files, Setup and Settingsfreakyjolly.com · Hands-on setup guide with GGUF sizes, ComfyUI node packs, and the licence catch excluding EU/UK/US/Korea.
- H3 squeezed to 13.7B: 768p video on a consumer GPUaimodeling.com · Technical breakdown of the three surgical cuts and NVIDIA's SOL-H3 validation measurements on GB200.
- Video-Rebirth/ComfyUI-HyperFlow-Flashgithub · 60 ★ · Official ComfyUI custom nodes for running HyperFlow H3 Flash workflows, GPL-3.0.
- NVlabs/Sanagithub · 9,816 ★ · NVIDIA's repo contains a validated models/minimax_h3/HyperFlow-Flash directory with SOL-H3 runtime measurements.
Learn more
First spotted on github: source.
More AI media tools
Hyperframes: write HTML, render video — HeyGen's open video engine for AI agentsOpen-source renderer from HeyGen that turns HTML, CSS and JavaScript into MP4 video, built so AI coding agents can…media · JEV 0.78Video Shotcraft: AI video skill for Claude Code & Codex that directs cinematic product promos in RemotionA skill package for Claude Code and Codex that turns a product brief into a finished promo video: the agent captures…media · JEV 0.71Lemo-Opuscar: 59 code-directed film styles for Claude CodeA toolbox of 59 film styles, each a style prompt plus a demo film made entirely in code by Claude Opus 5.5 with no…media · JEV 0.71SolidCraft: open-source parametric CAD in Rust that agents can drive over MCPSolidCraft is an open-source parametric 3D CAD application written in pure Rust (desktop, browser via WebAssembly…media · JEV 0.67Concat: free, open-source, local-first CapCut replacement scriptable by agentsLocal-first, cross-platform video editor that replaces CapCut, scriptable through a JSON-RPC API and MCP so coding…media · JEV 0.66Qwen-Image-2.1: 7B open image model that tops Nano Banana 2.0Alibaba's unified generation-and-editing image model: native 2K, native transparent-background output, up to 10…media · JEV 0.66
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.