Ant Bailing Open-Sources Ming-Image-0.1-Design for UI Generation
蚂蚁百灵开源 Ming-Image-0.1-Design 系列模型，UI 设计专项评测位列开源第一
An end-to-end design model that produces editable, layered assets directly addresses the multi-step fragility of chaining generic image models, and the transparent-layer output suits slide and web workflows. Developers building design or prototyping tools can test it free on OpenRouter and compare it against closed models in Artificial Analysis's UI/UX ranking. Weak spots in hand gestures and fine shadows are worth noting before production use.
Ant Group's Bailing team open-sourced the Ming-Image-0.
1-Design series on 23 September 2026, comprising two 6B-parameter models.The base model generates UI designs, infographics, posters and complete visual layouts from text, while the Layer variant splits images into independently editable transparent layers via a native RGBA VAE.
Ant also released Design Skill and PPT Skill to connect outputs to front-end pages and editable presentations.
In Artificial Analysis's UI/UX Design benchmark updated 18 September 2026, the model ranked first among open-source models with an Elo score of 1082 and layout win rates of 67.
4%.Free API access is available through OpenRouter for two weeks.
蚂蚁百灵开源 Ming-Image-0.1-Design 系列模型，UI 设计专项评测位列开源第一
IT之家 9 月 23 日消息，今日，蚂蚁百灵大模型正式开源 Ming-Image-0.1-Design 系列，包含两个参数量均为 6B 的模型：Ming-Image-0.1-Design，从文字需求生成 UI、信息图、海报和完整视觉设计；Ming-Image-0.1-Design-Layer，将创意图或者设计图拆成可独立编辑的透明图层。 IT之家注意到，蚂蚁百灵同时开源了 Design Skill 和 PPT Skill，将模型能力继续连接到前端页面与可编辑演示文稿。 据官方介绍，Ming-Image-0.1-Design 重点增强了四项能力： 支持 8K 长结构化提示词增强，将自然语言需求组织为文案、模块、布局和视觉风格； 优化标题、按钮、卡片和多区域信息的文字渲染与版式稳定性； 通过端到端生成一次性完成整体设计，统一配色、构图和素材风格； 创建原生 RGBA VAE，直接生成人物、商品、图标和装饰等透明背景素材。 相比通过代码多次调用通用生图模型，分别生成背景、插画和装饰，端到端生成拥有更完整的全局控制，能够减少素材之间色系冲突和画风不一致。 在 Artificial Analysis 于 2026 年 9 月 18 日更新的 UI / UX Design 专项评测中， Ming-Image-0.1-Design 位列开源模型第 1，Elo 得分为 1082。 模型在 UI / UX 场景中的版式胜率达到 67.4%，复杂构图达到 67.0%，文字渲染达到 66.7%。 官方称，Ming-Image-0.1-Design 系列仍有明确的能力边界。Design 目前最稳定的是排版和文字；但遇到复杂的人手动作、连续操作步骤，或者需要很精细的阴影、反光效果时，效果还不够稳定。在 Layer 模型中，分层不存在唯一答案。有人只需要背景、主体、文字三层，也有人希望将每段文字和每个图标分别拆开。大量叠加光效、特殊透明材质和复杂遮挡，也可能出现边缘残留或图层缺失。 为方便开发者快速体验，Ming-Image-0.1-Design 已在 OpenRouter 开放为期两周的免费 API 调用。
This report is categorized under AITOP's "Models" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Ming-Image-0.1-Design, Ant Group, Bailing, open-source models.
Source attribution: curated from IT之家. Cross-source corroboration: 2 independent sources have covered this development (including Artificial Analysis), confirming industry-wide signal strength.
- Artificial Analysis09-24 00:26Source
