Seedream 5.0 Pro adds multilingual layers
ByteDance released Seedream 5.0 Pro on July 8 with native generation in 10-plus languages, precision layer editing, and dense infographic creation.
ByteDance released Seedream 5.0 Pro on July 8 with native text generation in over 10 languages, including Chinese, English, French, German, Russian, Japanese, Korean, Spanish, and Arabic. A single image can also be separated into 10 or more independent editable layers.
The headline additions are in editing and layout control. Point, box, and lasso selection lets you relight, retexture, or replace specific objects while the rest of the composition stays intact. Intelligent layer separation exports each element as a transparent-channel PNG ready for compositing, cutting the back-and-forth with design software. A dedicated infographic mode handles dense layouts such as timelines, bar and pie charts, and multi-column data grids, and text rendering stays accurate across all supported languages.
The multilingual output goes beyond character support. The model adapts to the typographic conventions of each language, which matters for right-to-left scripts and languages with dense character sets where earlier image models typically fail on spacing and ligatures.
Availability is rolling out now. The model is live on the Volcano Ark platform; integration into Doubao and Jimeng follows. Dreamina offers free daily credits for individual users. Enterprise access is available through the BytePlus API.
For a comparison to where rival image generation stood before this release, see the Gemini Omni, Flash, and Nano overview. A practical workflow for business teams is in the AI images for business use guide.