The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.
Commits on master
Add pre attention and post input patches to qwen image model. (#12879)
9642e44 comfyanonymous committed 8 months ago
Allow model patches to have a cleanup function. (#12878)
3ad36d6 comfyanonymous committed 8 months ago
model_detection: deep clone pre edited edited weights (#12862)
a912809 rattus committed 8 months ago
Add a way for nodes to add pre attn patches to flux model. (#12861)
c4fb027 comfyanonymous committed 8 months ago
fix(manager): improve install guidance when comfyui-manager is not installed (#12810)
740d998 Dr.Lt.Data committed 8 months ago
Fix text encoder lora loading for wrapped models (#12852)
06f85e2 Jukka Seppänen committed 8 months ago
Import assets seeder later, print some package versions. (#12841)
e4b0bb8 comfyanonymous committed 8 months ago
refactor(assets): modular architecture + async two-phase scanner & background seeder (#12621)
29b24cb Luke Mino-Altherr committed 8 months ago
mm: reset_cast_buffers: sync compute stream before free (#12822)
bcf1a1f rattus committed 8 months ago
Don't run TE on cpu when dynamic vram enabled. (#12815)
d69d308 comfyanonymous committed 8 months ago
feat(api-nodes): add Gemini 3.1 Flash Lite model to LLM node (#12803)
34e55f0 Alexander Piskun committed 8 months ago
feat(api-nodes): add TencentSmartTopology node (#12741)
3b93d5d Alexander Piskun committed 8 months ago
feat: add Math Expression node with simpleeval evaluation (#12687)
e544c65 Dante committed 8 months ago
feat: add causal_fix parameter to add_keyframe_index and append_keyframe (#12797)
58017e8 Tavi Halperin committed 8 months ago
Add manual cast to LTX2 vocoder conv_transpose1d (#12795)
8befce5 Jukka Seppänen committed 8 months ago