Comfy-Org / ComfyUI

The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.

vllm-project / vllm

A high-throughput and memory-efficient inference and serving engine for LLMs

higgsfield-ai / higgsfield

Fault-tolerant, highly scalable GPU orchestration, and a machine learning framework designed for training models with billions to trillions of parameters