🟣 Patched-Metal / Specialized Runtime: Quantized DiT + Qwen3-VL Text Encoder β€” runs in stable-diffusion.cpp (sd-server / sd-cli, Metal 4) or ComfyUI.

Kenzan-MiniMax-H3-Image-20B-PinGrid

β˜• Support Local AI Development: Help me acquire a Mac mini 24GB/32GB to optimize and test local models up to 80B parameters. Donate via Ko-fi β†’

Kenzan-MiniMax-H3-Image-20B-PinGrid β€” "20B DiT + 25B Qwen3-VL Text Encoder + 1.1GB VAE, Running Locally on 16 GB RAM." 37.47 GiB BF16 DiT compressed 4.46Γ— to 8.40 GiB, int8_convrot 50-layer Qwen3-VL-32B unpacked and Pin-Grid quantized to 10.46 GiB, and native Apple Metal (stable-diffusion.cpp) delivering ~7.2s/step at 1024Γ—1024 with zero OOMs.

Overview

This repository provides a complete, 16 GB Unified Memory-ready GGUF release of MiniMaxAI/MiniMax-H3-Image (from Comfy-Org/MiniMax-H3-Image) tuned for Apple Silicon (Mac mini M6 16 GB) and compatible with both stable-diffusion.cpp (sd-cli / sd-server with Metal) and ComfyUI + ComfyUI-GGUF.

Included Files

Component File Size Compression / Precision
Diffusion Transformer (20.11B, 50 blocks) Kenzan-MiniMax-H3-Image-20B-PinGrid-Q3_K-M6-16G.gguf 8.40 GiB Pin-Grid Q3_K / Q4_0 / Q5_K / Q2_K (F32 norms & adaln) β€” 4.46Γ— vs BF16 (37.47 GiB)
Text/Vision Encoder (Qwen3-VL-32B L50) Kenzan-Qwen3VL-32B-L50-MiniMaxH3-PinGrid-Q2_K-M6-16G.gguf 10.46 GiB Dequantized from int8_convrot (25.28 GiB), Pin-Grid Q2_K/Q3_K/Q4_0 + Q8_0 visual.* for Image Edit
Image VAE (MiniMaxH3ImageVAE) minimax_h3_image_vae_fp16.safetensors 1.09 GiB Original lossless FP16
Quantization Recipes recipes/convert_dit_to_gguf.py, recipes/convert_te_to_gguf.py β€” Exact streaming GGUF conversion scripts

Pin-Grid Architecture

  • 20B DiT (Kenzan-MiniMax-H3-Image-20B-PinGrid-Q3_K-M6-16G.gguf):
    • token_refiner.*, final_layer.*: Q5_K
    • Entry & tail blocks blocks.(0,1,48,49) + blocks.(40..47).attn.* + peak mlp.fc2 (8, 16, 24, 32, 40): Q4_0
    • Mid-late blocks.(20..47): Q3_K
    • Early MLP blocks.(2..19): Q2_K
    • adaln_t_table, adaln_proj, patch_proj, norms: F32
  • 50-Layer Qwen3-VL-32B Text Encoder (Kenzan-Qwen3VL-32B-L50-MiniMaxH3-PinGrid-Q2_K-M6-16G.gguf):
    • visual.* (for Image Edit): Q8_0
    • model.embed_tokens: Q5_K
    • Entry & exit layers 0, 48, 49 + k_proj/v_proj + peak mlp.down_proj (8, 16, 24, 32, 40): Q4_0
    • Mid-late layers.(20..47): Q3_K, early MLP layers.(1..19): Q2_K

Verified Benchmarks (Apple Mac mini M6, 16 GB Unified Memory, 1024Γ—1024, 6 steps euler/simple)

Backend DiT Step Speed VAE Decode Peak RSS Notes
ComfyUI + ComfyUI-GGUF (PyTorch MPS) 13.27 s/step 5.36 s 11.88 GiB Per-layer runtime dequant to FP16
stable-diffusion.cpp (sd-cli / sd-server, Metal 4) 7.20 – 8.24 s/step 2.43 – 2.59 s Fits 16 GB (--offload-to-cpu) 1.8Γ— faster DiT, 2.1Γ— faster VAE

Quick Start (stable-diffusion.cpp on Apple Silicon 16 GB)

CLI (sd-cli)

./sd-cli   --diffusion-model Kenzan-MiniMax-H3-Image-20B-PinGrid-Q3_K-M6-16G.gguf   --llm Kenzan-Qwen3VL-32B-L50-MiniMaxH3-PinGrid-Q2_K-M6-16G.gguf   --vae minimax_h3_image_vae_fp16.safetensors   --offload-to-cpu --flash-attn on   -W 1024 -H 1024 --steps 6 --cfg-scale 1.0   --sampling-method euler --scheduler simple   -p "Your prompt here" -o output.png

Persistent Server (sd-server, eliminates cold-start weight loading)

./sd-server --listen-ip 0.0.0.0 --listen-port 8189   --diffusion-model Kenzan-MiniMax-H3-Image-20B-PinGrid-Q3_K-M6-16G.gguf   --llm Kenzan-Qwen3VL-32B-L50-MiniMaxH3-PinGrid-Q2_K-M6-16G.gguf   --vae minimax_h3_image_vae_fp16.safetensors   --offload-to-cpu --flash-attn on

Acknowledgements & Inspiration


Support the Project & Funding Goals

All Kenzan weights and quantization recipes are released openly and free of charge for the local AI community.

β˜• Donate via Ko-fi β†’

What Your Donations Directly Fund

  1. Upgrading to a Mac mini 24 GB / 32 GB (Apple Silicon): Breaking past the 16 GB ceiling to quantize and benchmark 70B–80B models.
  2. Cloud GPU Compute for Calibration Sweeps: Covering VRAM hours for imatrix generation and layer sensitivity sweeps.
  3. Open-Source Metal Kernels & Tooling: Funding continued development of llama.cpp / stable-diffusion.cpp Metal optimizations.
Downloads last month
261
GGUF
Model size
20B params
Architecture
minimax_h3
Hardware compatibility
Log In to add your hardware

2-bit

Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Model tree for kabcnd/Kenzan-MiniMax-H3-Image-20B-PinGrid-GGUF

Quantized
(5)
this model

Collection including kabcnd/Kenzan-MiniMax-H3-Image-20B-PinGrid-GGUF