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
Fix issue with parameters on root model object. (#12216)
021ba20 comfyanonymous committed 9 months ago
dynamic_vram: respect argument cast dtypes in non-comfy weights (#12209)
794d05b rattus committed 9 months ago
fix: improve error message when node type is missing (#12194)
32621c6 Christian Byrne committed 9 months ago
Reduce RAM usage, fix VRAM OOMs, and fix Windows shared memory spilling with adaptive model loading (#11845)
f8acd9c rattus committed 9 months ago
KV cache implementation for using llama models for text generation. (#12195)
873de5f comfyanonymous committed 9 months ago
Send is_input_list on v1 and v3 schema to frontend (#12188)
aa6f7a8 Jedrzej Kosinski committed 9 months ago
feat(api-nodes): add Q3 models and support for Extend and MultiFrame Vidu endpoints (#12175)
6e469a3 Alexander Piskun committed 9 months ago
feat(api-nodes): add RecraftCreateStyleNode node (#12055)
0167653 Alexander Piskun committed 9 months ago
Remove NodeInfoV3-related code; we are almost 100% guaranteed to stick with NodeInfoV1 for the foreseable future (#12147)
0a79937 Jedrzej Kosinski committed 9 months ago
Make empty hunyuan latent 1.0 work with the 1.5 model. (#12171)
bbe2c13 comfyanonymous committed 9 months ago
Update Python 3.14 compatibility notes in README (#12127)
4f5bd39 comfyanonymous committed 9 months ago