Skip to content

EthenEthenEthen

LoginPlatformTry Ethen

Open Source Model Profile · HiDream-ai

HiDream-I1-Fast

HiDream-I1-Fast is a 17B-parameter distilled text-to-image model from HiDream-ai. According to the model card, it is the fast distilled variant of the HiDream-I1 sparse diffusion transformer family.

Publisher
HiDream-ai
Task
text-to-image
Model type
Unknown
License
mit
Library
diffusers
Publication status
Accepted · not indexed

Model overview

HiDream-I1-Fast is published by HiDream-ai as a text-to-image generative model. Captured Safetensors metadata reports 17,105,733,184 parameters, about 17.11B. According to the model card, HiDream-I1 is an open-source image foundation model with 17B parameters that generates images within seconds, and the inference script exposes full, dev, and fast model types.

Recorded capabilities

Distilled fast variant

According to the model card, the fast model type runs the distilled inference path alongside full and dev options.

Publisher-reported benchmark tables

According to the model card, HiDream-I1 reports 85.89 overall on DPG-Bench, 0.83 overall on GenEval, and 33.82 averaged on HPSv2.1.

Sparse diffusion transformer

According to the model card's project update, the family uses a sparse diffusion transformer described in its technical report.

Component license split

According to the model card, Transformers are MIT-licensed while the VAE comes from FLUX.1 schnell and text encoders carry their own terms.

Use cases in the source record

  • Fast text-to-image generation using the card's fast-mode inference script and Gradio demo for interactive use.
  • Prompt-following evaluation studies referencing the card's GenEval, DPG-Bench, and HPSv2.1 tables for the HiDream-I1 family.

Limitations and unknowns

  • No independently measured evaluation results were extracted; benchmark tables are publisher-reported card claims for the HiDream-I1 family.
  • No training data, compute, or hyperparameter detail was extracted from this record.
  • Component licenses beyond the Transformer models require following the VAE and text-encoder terms stated in the card.
  • Provider state is historical snapshot data, not independently refreshed current availability.

Source and provenance

Source: HiDream-ai/HiDream-I1-Fast

Captured: Unknown. Processed: 2026-09-07T19:35:06.299989+00:00.

HiDream-I1 is a new open-source image generative foundation model with 17B parameters that achieves state-of-the-art image generation quality within seconds. For more features and to experience the full capabilities of our product, please visit https://vivago.ai/ . Project Updates 📝 May 28, 2025 : We've released our technical report HiDream-I1: A High-Efficient Image Generative Foundation Model with Sparse Diffusion Transformer . 🚀 April 28, 2025 : We've open-sourced the instruction-based-image-editing model HiDream-E1-Full . Experience at https://huggingface.co/spaces/HiDream-ai/HiDream-E1-Full !. Key Features ✨ Superior Image Qu…

F001F002F003F004F005F006F008F009F011F014F015F016F017F018F019F020