Distilled 8-step generation
According to the model card, ERNIE-Image-Turbo is distilled from ERNIE-Image and optimized by DMD and RL for faster, higher-aesthetic generation in 8 steps.
Open Source Model Profile · baidu
ERNIE-Image-Turbo
ERNIE-Image-Turbo is an 8.03B-parameter text-to-image model from Baidu. According to the model card, it is a distilled DiT variant designed for 8-step generation.
- Publisher
- baidu
- Task
- text-to-image
- Model type
- Unknown
- License
- apache-2.0
- Library
- diffusers
- Publication status
- Accepted · not indexed
Model overview
ERNIE-Image-Turbo is published by baidu as a text-to-image model. Safetensors metadata reports 8,033,490,048 parameters, about 8.03B. According to the model card, it is the distilled release of ERNIE-Image in the single-stream Diffusion Transformer family for fast generation with strong fidelity.
Recorded capabilities
Text rendering and instruction focus
According to the model card, the model performs well on dense, layout-sensitive text and follows complex prompts with multiple objects and relationships.
Documented diffusers workflow
According to the model card, generation uses ErnieImagePipeline at guidance 1.0 for 8 steps, with install and resolution guidance and an SGLang serving example.
Publisher-reported LongTextBench table
According to the model card, the card publishes LongTextBench English and Chinese comparisons that include ERNIE-Image-Turbo variants alongside other image models.
Use cases in the source record
- Poster, comic, multi-panel, and infographic generation where text rendering and structured layouts matter, as described in the card.
- Latency-sensitive 8-step text-to-image generation using the card's recommended resolutions, guidance 1.0, and diffusers setup.
Limitations and unknowns
- No independently measured evaluation results were extracted; benchmark tables are publisher-reported card claims.
- No training data, compute, or hyperparameter detail beyond DMD and RL naming was extracted.
- Provider state is historical snapshot data, not independently refreshed current availability.
- Download and ecosystem figures in the card are captured page text, not Ethen-verified current counts.
Source and provenance
Source: baidu/ERNIE-Image-Turbo
Captured: Unknown. Processed: 2026-09-07T19:35:54.465163+00:00.
ERNIE-Image-Turbo 🤗 ERNIE-Image | 🤗 ERNIE-Image-Turbo | 🤖 ERNIE-Image | 🤖 ERNIE-Image-Turbo 🖥️ Huggingface Demo1 | 🖥️ Huggingface Demo2(ZeroGPU) | 🖥️ AI Studio Demo Github | 📖 Blog | 🖼️ Art Gallery 💬 WeChat(微信) | 🫨 Discord | 🏷️ X ERNIE-Image-Turbo is an open text-to-image generation model developed by the ERNIE-Image team at Baidu. It is the distilled release of ERNIE-Image, built on the same single-stream Diffusion Transformer (DiT) family and designed for fast generation with strong fidelity in only 8 inference steps. The model retains strong controllability in practical generation scenarios where accurate content real…
F001F002F003F004F005F006F008F009F010F011F012F014F015F016F018F020