Attempt to build sage attention 2.2+ it speeds up MiniMax significantly.

Tested on rtx5070ti.

Small info: bypass PARAMS / MAIN every time Ollama is active

Diffusion model:

• https://huggingface.co/Comfy-Org/MiniMax-H3/blob/main/diffusion_models/minimax_h3_ref2va_pruned_int8_convrot.safetensors

Text encoder:

• https://huggingface.co/sakamakismile/Qwen3-VL-32B-Heretic-MiniMax-H3-NVFP4/blob/main/qwen3vl_32b_heretic_minimax_h3_nvfp4.safetensors

VAEs:

• https://huggingface.co/Comfy-Org/MiniMax-H3/tree/main/vae

Ollama LLM (gemma 4 12B it) for prompt helper:

• https://huggingface.co/mradermacher/gemma-4-12B-it-uncensored-heretic-GGUF/blob/main/gemma-4-12B-it-uncensored-heretic.Q8_0.gguf

You can add as many references as needed..

DOWN BELOW IS OLD VERSION CONTENT.

DISABLE comfyui_smznodes if you have this extension!!!

(It breaks any LTXV workflow.)

Diffuser model (pick one):

• https://huggingface.co/vantagewithai/LTX2.3-10Eros-1.4-GGUF/blob/main/10Eros_v1.4-Q4_K_S.gguf

• https://huggingface.co/unsloth/LTX-2.3-GGUF/blob/main/ltx-2.3-22b-dev-Q4_K_S.gguf

Text encoders (both):

• https://huggingface.co/Comfy-Org/ltx-2/blob/main/split_files/text_encoders/gemma_3_12B_it_fp8_scaled.safetensors

• https://huggingface.co/unsloth/LTX-2.3-GGUF/blob/main/text_encoders/ltx-2.3-22b-dev_embeddings_connectors.safetensors

VAEs:

• https://huggingface.co/Kijai/LTX2.3_comfy/tree/main/vae

Latent upscaler:

• https://huggingface.co/Lightricks/LTX-2.3/blob/main/ltx-2.3-spatial-upscaler-x2-1.1.safetensors

LoRA (distilled LoRA is needed, do not disable):

• https://huggingface.co/Kijai/LTX2.3_comfy/blob/main/loras/ltx-2.3-22b-distilled-lora-dynamic_fro09_avg_rank_105_bf16.safetensors

Enjoy!