CoolFace
Modelpublic

gabriellarson/UIGEN-T3.1-8B-Full-Demo-No-Reasoning-GGUF

sourceHugging Faceapache-2.0updated 1y agoView on Hugging Face
0likes884downloads
Model Card

<img src="https://cdn-uploads.huggingface.co/production/uploads/67db34a5e7f1d129b294e2af/ZTVLfTQv_daPXu5q8Zk5s.webp" alt="UIGEN-T3 UI Screenshot 1" width="300">

UIGEN-X-8B — Streamlined UI Generation

Tesslate's optimized UI model, built for direct and efficient code generation. Optimized for popular web develpment stacks such as:

React Ecosystem

  • —React + Tailwind CSS: shadcn/ui, NextUI, DaisyUI, Headless UI, Flowbite React, Tremor, Park UI, Catalyst
  • —React Standalone: Material-UI (MUI), Ant Design (AntD), Chakra UI, Mantine, PrimeReact
  • —Meta Frameworks: Next.js, Remix, Vite, Create React App

Vue.js Ecosystem

  • —Component Libraries: Vuetify, PrimeVue, Tailwind CSS direct integration
  • —Meta Framework: Nuxt.js
  • —Styling: Material Design, comprehensive component suites

Angular Enterprise

  • —Libraries: Angular Material, PrimeNG, ng-bootstrap
  • —Styling: SCSS, CSS, Bootstrap integration
  • —Framework: Angular CLI with enterprise-grade state management

Supported Frameworks: React, Vue.js, Angular, Svelte, Vanilla JS, and more

Styling Engines: Tailwind CSS, Material-UI, Emotion, Bootstrap, SCSS, CSS-in-JS

Component Libraries: 50+ including shadcn/ui, NextUI, Vuetify, Angular Material

Demos

<img src="https://cdn-uploads.huggingface.co/production/uploads/67db34a5e7f1d129b294e2af/ILHx-xcn18cyDLX5a63xV.png" alt="UIGEN-T3 UI Screenshot 1" width="800"> <img src="https://cdn-uploads.huggingface.co/production/uploads/67db34a5e7f1d129b294e2af/A-zKo1J4HYftjiOjq_GB4.png" alt="UIGEN-T3 UI Screenshot 1" width="800">

Explore New UI generations: 📂 https://uigenoutput.tesslate.com

Join our Discord: https://discord.gg/GNbWAeJ4 Our Website: https://tesslate.com

Quick Information

  • —UI generation model built on Qwen3 architecture
  • —No-reasoning variant for faster, direct code generation
  • —Supports both components and full web pages
  • —116 web tech stacks supported across React, Vue, Angular, Svelte, and more
  • —Most popular stacks: React+Tailwind (27 variants), Material-UI ecosystem, Vue+Nuxt
  • —Streamlined output: Pure HTML and Tailwind CSS without reasoning overhead
  • —Powered by UIGenEval, a first-of-its-kind benchmark for UI generation
  • —Released under Apache 2.0 license for research and commercial use

Model Details

  • —Base Model: unsloth/Qwen3-8B-unsloth-bnb-4bit
  • —Reasoning Style: No reasoning (direct generation)
  • —Tokenizer: Qwen default, with design token headers
  • —Output: Components + Full pages (with <html>, <head>)
  • —Images: User-supplied or placehold.co – no images in the dataset due to licensing concerns
  • —License: Apache 2.0 (permissive commercial use)

Architecture System

UIGEN-T3.1-8B-Full-Demo-No-Reasoning was trained using a streamlined architecture optimized for direct code generation.

Unlike hybrid reasoning variants, this model focuses on:

  • —Direct code generation → Fast, efficient HTML/CSS output without reasoning blocks
  • —Consistent formatting → Standardized structure for production-ready code
  • —Optimized inference → Reduced latency compared to reasoning-enabled models

Outputs include design tokens at the top of each generation for easier site-wide customization.

Inference Parameters

For optimal performance with this no-reasoning variant:

ParameterValue
Temperature0.6
Top P0.95
Top K20
Max Tokens40k+

Evaluation: UIGenEval Framework

UIGenEval is our internal evaluation suite, designed to bridge the gap between creative output and quality assurance. (Learn more in our upcoming paper: "UIGenEval: Bridging the Evaluation Gap in AI-Driven UI Generation" - August, 2025)

UIGenEval evaluates models across four pillars:

  1. 1.Technical Quality — Clean HTML, CSS structure, semantic accuracy
  2. 2.Prompt Adherence — Feature completeness and fidelity to instructions
  3. 3.Interaction Behavior — Dynamic logic hooks and functional interactivity
  4. 4.Responsive Design — Multi-viewport performance via Lighthouse, Axe-core, and custom scripts

This comprehensive framework directly informs our GRPO reward functions for the next release.

Example Prompts to Try

  • —Design a multi-vendor marketplace homepage with live product carousels, seller spotlight cards, and a dynamic filters sidebar using React + Next.js + Tailwind CSS + shadcn/ui. Include loading skeletons and dark mode support.
  • —Generate a subscription management dashboard for a SaaS product (with plan comparisons, usage analytics, billing history, and upgrade modal) in Vue + Nuxt.js + Vuetify. The dashboard should support both admin and user views.
  • —Build an enterprise-grade analytics dashboard with filterable data tables, export buttons, real-time chart widgets (Tremor + Recharts), and user management features. Use React + Next.js + Tailwind CSS + Tremor. Add role-based sidebar navigation and mobile responsiveness.
  • —Create a healthcare metrics dashboard with multi-level drill-down, interactive graphs, and secure login using Angular + Angular Material + SCSS. Support light/dark themes and modular cards.
  • —Design a customizable headless CMS editor panel (fields: text, images, rich embeds, SEO settings) in SvelteKit + Tailwind CSS + Skeleton UI. Embed drag-and-drop section reordering and inline editing.

Use Cases

Use CaseDescription
Rapid PrototypingQuick UI mockups without reasoning delays
Production TemplatesClean, ready-to-use HTML/Tailwind components
Batch GenerationEfficient processing of multiple UI requests
API IntegrationStreamlined for REST API and automated workflows
Educational ToolsTeaching HTML/CSS without complex reasoning explanations

Limitations

  • —No reasoning explanations — Pure code output without design rationale
  • —Limited complex logic — Best suited for UI structure, not interactive features

Technical Requirements

  • —GPU: ≥8GB VRAM for 8B inference
  • —Libraries: transformers, torch, unsloth
  • —Python: 3.8+
  • —Compatible with Hugging Face inference APIs and local generation pipelines

Community & Contribution

  • —Join our Discord: https://discord.gg/GNbWAeJ4
  • —Chat about AI, design, or model training
  • —Want to contribute UIs or feedback? Let's talk!

Citation

bibtex
@misc{tesslate_UIGEN-X,
  title={UIGEN-X: Efficient UI Generation model built on Qwen3},
  author={Tesslate Team},
  year={2025},
  publisher={Tesslate},
  note={Non-commercial Research License},
  url={https://huggingface.co/tesslate/UIGEN-T3}
}