Qwen3-VL-32B model optimized for ComfyUI with INT8 quantization

A modified version of the Qwen3-VL-32B model, named "Ultra Uncensored Heretic", has been optimized for use with ComfyUI. This version, "MiniMax-H3", features INT8 quantization and ConvRot layers to reduce its size, making it suitable for GPUs with 32GB of VRAM like the RTX 5090. The release includes two safetensors files: a primary conditioning encoder and an optional prompt-enhancement tail for improved generation. AI

IMPACT Enables running larger, more capable models on consumer hardware by optimizing for VRAM usage.

RANK_REASON This is a release of a modified, quantized model for a specific UI, not a new frontier model release from a major lab.

Read on Hugging Face Trending Models →

• ComfyUI
• ConvRot
• ethanfel/Qwen3-VL-32B-Ultra-Heretic-MiniMax-H3-ComfyUI-INT8-ConvRot
• Int8
• llmfan46/Qwen3-VL-32B-Instruct-ultra-uncensored-heretic
• Qwen3-VL-32B Ultra Uncensored Heretic
• RTX 5090

• model release
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →