This will take your prompt, enhance it and optimize it (using LLMs) for the T5XXL, CLIP-L, and CLIP-G text encoders that SD3.x uses.

SD3.5 link: https://huggingface.co/stabilityai/stable-diffusion-3.5-large