ERNIE-Image is a text-to-image diffusion model hosted on Hugging Face under the baidu organization. It generates images from textual prompts and belongs to the class of diffusion-based text-to-image models.
The model is provided with an ErnieImagePipeline implementation and is distributed in Safetensors format. It can be loaded through the Diffusers library using a few lines of Python code that specify a torch bfloat16 data type and a CUDA device map. Example usage includes creating a pipeline from the repository identifier and passing a prompt such as "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" to produce an output image.
Integration is available via several inference providers, Colab and Kaggle notebooks, and local applications including Draw Things and DiffusionBee. The repository lists benchmark results on GENEval, OneIG-EN, OneIG-ZH, and LongTextBench. A lighter variant called ERNIE-Image-Turbo is referenced alongside the main model.
ERNIE-Image is released under the Apache 2.0 license. The page indicates 3.24k likes and 664 followers for the repository.
ERNIE Image is an Image generation product. It focuses on generating detailed images from text descriptions using a powerful diffusion model. It is built as an open-source project for developers. ERNIE Image is open source under the Apache-2.0 license. It runs on the web, the command line, and API.
It is developed by Baidu (China), and the product first shipped in 2026. Development happens publicly on GitHub with 493 stars. Key capabilities include text-to-Image, Diffusion Model, and High Resolution.
Latest indexed changes and source events
baidu/ERNIE-Image verified by the PulseGate indexer
Other apps tracked under the same category.
