TensorRT-Optimized Stable Diffusion 3.5 Models Deliver 2X Faster Performance and 40% Less Memory on NVIDIA RTX GPUs

Stability AI and NVIDIA released TensorRT-optimized Stable Diffusion 3.5 models for RTX GPUs.

The models use FP8 to speed up image generation and reduce VRAM usage.

Stable Diffusion 3.5 Large is up to 2.3x faster with 40% less memory, while Medium is up to 1.7x faster.

The models are available for commercial and non-commercial use via Hugging Face and NVIDIA GitHub.