AesCode-32B

Use it to create editable slides, posters, dashboards, and reports from content prompts, optionally guided by a reference image.

Availability

Companion model available on Hugging Face

Pricing model

Apache 2.0 license

Deployment

Load locally with Transformers; serve with vLLM

Privacy

Candidate HTML is scored in a sandboxed browser with external requests blocked

Licence

Apache 2.0

• AesCode-32B generates visual artifacts such as slides, posters, and dashboards as HTML/CSS. ^[1]

  AesCode generates information-rich visual artifacts such as slides, posters, and

• The model can use a reference image as guidance for layout and style. ^[1]

  detailed content prompt and, when available, a reference image for layout and style

• The usage instructions require transformers>=4.57 and estimate roughly 65 GB of accelerator memory for the parameters. ^[1]
• The model emits complete HTML documents, with tables using HTML table structures and charts using ECharts specifications. ^[1]
• The model is released under the Apache 2.0 license. ^{[1] [2]}

  Released under the Apache 2.0 license, following the Qwen3-VL backbone.

• AesCode-32B generates visual artifacts as structured HTML/CSS. ^[1]
• A reference image is optional at inference. ^[1]
• The model follows the Qwen3-VL chat interface. ^[1]
• The model can be served with vLLM. ^[1]

• Turn a prompt into an editable slide, poster, or dashboard Weekly brief 2026-W41
• Generate editable slides and dashboards from a prompt Daily brief 2026-10-11