HiDream-I1-Full is the top version of HiDream-ai’s open image model released April 2025. It runs on a 17 billion parameter Diffusion Transformer with dynamic expert mixing. Claims say it beats others in speed and image quality.
It runs 50 steps needing about 27 GB of VRAM. Dev and Fast versions use 28 and 16 steps. It’s open source under MIT with full commercial rights.
It scores well in GenEval DPG-Bench and HPS v2.1 with 33.8 average accuracy.
It’s flexible with styles like photoreal cartoon art and anime. It uses four text encoders: OpenCLIP CLIP T5-XXL and Llama 3.1 8B Instruct.
Despite its size it pushes out sharp results fast.
If you'd like to access this model, you can explore the following possibilities:
Use our video cost calculator to compare prices between platforms offering HiDream I1 Full model.
For locally hosted models, see description and additional links at the bottom for versions, repos and tutorials.
