zen-designer-235b-a22b-instruct

Design-specialized vision-language model for UI/UX, visual creation, and multimodal design tasks. Instruction-tuned variant.

Fine-tuned from Qwen/Qwen3-VL-235B-A22B-Instruct (apache-2.0, Alibaba Qwen) with Hanzo identity + agentic-data training + abliteration. Not trained from scratch.

Specs

Property	Value
Parameters	235B total, 22B active (sparse MoE)
Architecture	Qwen3-VL MoE (Qwen3VLMoeForConditionalGeneration)
Modality	Vision-language (image understanding + text)
Context	131K tokens

Design Capabilities

• UI/UX: dashboard, mobile, and web layouts; component and design systems
• Visual analysis: critique and improve existing interfaces from an image
• Technical export: CSS, React components, design tokens

Quick Start

from transformers import AutoModelForImageTextToText, AutoProcessor

model = AutoModelForImageTextToText.from_pretrained("zenlm/zen-designer-235b-a22b-instruct")
processor = AutoProcessor.from_pretrained("zenlm/zen-designer-235b-a22b-instruct")

messages = [{"role": "user", "content": [{"type": "text", "text": "Design a modern analytics dashboard."}]}]
inputs = processor.apply_chat_template(messages, add_generation_prompt=True, tokenize=True, return_tensors="pt")
out = model.generate(**inputs, max_new_tokens=512)
print(processor.decode(out[0], skip_special_tokens=True))

License

apache-2.0. Upstream: Qwen/Qwen3-VL-235B-A22B-Instruct by Alibaba Qwen (apache-2.0).

---

Built by Hanzo AI × Zoo Labs Foundation