All issues

Search results

SageAttention 2.2.0 registers its kernels as torch custom ops, so _resolve_qattn returns None (AttributeError mid-sampling)

Status: Open.

#746 In kijai/ComfyUI-KJNodes;

· leandrokoiti opened on Aug 26, 2026

Feature Request: can "minimax_nodes.py" include layer-based attn strength filtering for AUDIO in LoRAs like "LTX2LoraLoaderAdvanced"?

Status: Open.

#745 In kijai/ComfyUI-KJNodes;

· 808charlie opened on Aug 26, 2026

win10-CUDA-12.6/13.0 Python-3.12.10/3.13.12：Successfully installed sageattention-2.2.0

Status: Open.

#743 In kijai/ComfyUI-KJNodes;

· Dao28 opened on Aug 25, 2026

[FEATURE REQUEST] 4DAnyone

Status: Open.

#742 In kijai/ComfyUI-KJNodes;

· alexfredo opened on Aug 21, 2026

Node's tooltip to have consumer/noob friendly info?

Status: Open.

#741 In kijai/ComfyUI-KJNodes;

· CherrySkyArt opened on Aug 21, 2026

Feature Request: Streamed QKV + Near-Lossless 4-bit K/V Compression for MiniMax H3 Low-VRAM Attention

Status: Open.

#738 In kijai/ComfyUI-KJNodes;

· eliowang6 opened on Aug 16, 2026

[FEATURE] Minimax H3 LoRa Loader Advanced

Status: Open.

#737 In kijai/ComfyUI-KJNodes;

· GZ-572 opened on Aug 14, 2026

MiniMax H3 Memory Efficient SageAttention Patch fails on newer SageAttention wheels because _qattn_sm89 resolves to None

Status: Open.

#736 In kijai/ComfyUI-KJNodes;

· QQWWLOOK opened on Aug 14, 2026

Feature/business idea: pay-once tiered licensing option

Status: Open.

#735 In kijai/ComfyUI-KJNodes;

· kojiyan opened on Aug 12, 2026

Is teah3 no longer available in ModelPreviewOverrideKJ? It was working fine a few days ago.

Status: Open.

#733 In kijai/ComfyUI-KJNodes;

· wenchengxiang opened on Aug 8, 2026

Mr. KJ, we need those genius‑level magic nodes like LTX. :)

Status: Open.

#732 In kijai/ComfyUI-KJNodes;

· ptmaster opened on Aug 8, 2026

Context Windows node for MiniMax H3?

Status: Open.

#731 In kijai/ComfyUI-KJNodes;

· Kvento opened on Aug 8, 2026