字节跳动发布图像创作模型Seedream 5.5

1. 10:25
   AITNT · AI头条

   豆包 Seedream 5.5 实测：支持百轮稳定局部编辑与图层拆分

   字节跳动旗下豆包正式发布新一代图像生成模型 Seedream 5.5 并上线豆包及豆包工作，实测展现了其在多轮编辑与管线整合上的能力。该模型重点提升了多轮编辑稳定性，可实现调整 100 轮非编辑区像素级保持，并支持原生 RGBA 透明通道与最多约 100 层的图层拆分。在豆包工作中，模型可作为 Skill 结合飞书群聊讨论与文档自动梳理需求，直接串联海报制作与网页物料搭建。

2. 09:22
   AITNT · AI头条

   字节跳动发布 Seedream 5.5 图像创作模型：支持 100 轮连续编辑与原生透明素材

   字节跳动 Seed 团队正式发布新一代图像创作模型 Seedream 5.5，重点强化了最多 100 轮的多轮编辑稳定性、原生 RGBA 透明素材生成以及复杂文字与质感渲染。该模型单张图片生成耗时约 5-10 秒，已在豆包、豆包工作上线，并陆续开放至即梦AI与火山方舟。API 采用输入图与输出图分开计费且文本 Token 免费，2K 方图编辑单张成本约 0.28 元。

3. 02:50
   The DecoderSelected

   ByteDance Releases Seedream 5.5 Image Generation Model

   ByteDance Seed has introduced Seedream 5.5, an image generation model tailored for professional production workflows. It supports up to 100 rounds of multi-turn editing while maintaining pixel-level stability and zero color shift in unedited areas. The model also supports native RGBA images with transparent backgrounds and brings improvements in lighting, textures, skin realism, and dense, fine-text rendering. It is gradually rolling out across Jimeng, Doubao, Doubao Work, and Volcano Engine Ark, with a Pro version planned for later release.

1. 20:01
   爱范儿 · AI 筛选Selected

   Hands-on with Doubao Seedream 5.5: Focused on Multi-Turn Precise Editing and Layer Separation, Integrated into Design Workflows

   The next-generation image generation model Doubao Seedream 5.5 has officially launched and is rolling out across Doubao and Doubao Work, with a focus on enhancing multi-turn localized editing, layer separation, and native alpha channel capabilities. Hands-on testing shows that it achieves pixel-level preservation in unedited areas across multi-turn edits, supports decomposing a single image into roughly 100 element layers, and directly outputs RGBA transparent PNG assets. Within Doubao Work, the model can link Feishu group chat requests with downstream asset production via skills, integrating AI image generation into a complete design delivery pipeline.

2. 17:35
   IT之家 · AI 筛选

   字节跳动发布图像创作模型 Seedream 5.5

   字节跳动 Seed 团队发布新一代图像创作模型 Seedream 5.5，支持多达 100 轮连续编辑且非编辑区域保持像素级一致，并原生支持输出带 Alpha 通道的 RGBA 透明素材。

3. 17:19
   爱范儿 · AI 筛选Selected

   ByteDance Releases Seedream 5.5 Image Generation Model

   ByteDance has launched its next-generation image creation model, Seedream 5.5, highlighting pixel-level localized editing, consistency across hundreds of continuous revision rounds, and native RGBA transparency asset generation. Tests indicate that its severe text-rendering failure rate in complex bilingual (Chinese and English) and fine-print typography has dropped to 0.4%, with API pricing roughly one-third that of GPT Images 2.6. The model has already been integrated into Doubao Work and Jimeng, with Volcano Engine set to launch API services this week.

Current heat 30·Comparable peak 31（Nov 10 10:00）·Comparable change, last 24 hours–

Trends compare only the same parties observed continuously, so the scope can be narrower than current heat counts. Move the pointer or tap the chart to see hourly heat; use the left and right arrow keys on a keyboard.