Ant Group · image_gen
Ming-Image-0.1-Design
Ming-Image-0.1-Design is an MIT-licensed 6B text-to-image checkpoint for UI layouts, infographics and text-rich posters. It supports transparent RGBA output; the official reference setup uses an 80 GiB CUDA GPU.
Downloadable checkpoint · MIT
Official sources reviewed 2026-10-04. This page documents access and specifications; it does not report a model benchmark.
Task and deployment contract
Download the Design checkpoint and use the companion Ming-Image repository with the text-to-image task.
The model generates a complete visual design. It does not produce editable HTML, CSS or a structured slide deck by itself. Image-to-editable-PPT tooling is a separate downstream workflow.
The reference settings are BF16, 12 sampling steps and CFG 1.0. One CUDA GPU with 80 GiB VRAM is the validated configuration; this does not establish a minimum hardware requirement or measured latency.
For native transparency, follow the repository's recommended RGBA prompt prefixes and inspect the output alpha channel. A checkerboard in a preview is not an image layer.
The model card and checkpoint LICENSE specify MIT. No current hosted rate has been verified for this directory; downloadable weights and provider billing are separate contracts.
Billing and limits
No current provider rate for this exact configuration is recorded here. Open weights still require compute and operations; an unlisted rate does not mean a hosted API is free.
Image limits use resolution buckets, not a chat-token context window. The documented buckets are 1024 and 2048; 2048 x 2048 is recommended.
Fast Pivot — Access hundreds of AI models from a single API — competitive pricing, reliable uptime, enterprise-grade infrastructure.