🚀 Welcome to the UIGEN Playground 🚀

Specialized UI based reasoning

Model by TesslateAI | View on Hugging Face | Running with 8-bit Quantization | Streaming Output

Interact with Tesslate/UIGEN-T2-7B. UIGEN-T2 specializes in frontend development, leveraging advanced reasoning to autonomously generate well-structured components. This demo uses 8-bit quantization via bitsandbytes for reduced memory footprint. Flash Attention 2 is enabled if available. Output is streamed token-by-token.
0 2
64 10000
1 200
0.05 1
1 2
0 0.5

Model Configuration

Model Type: qwen2 Architecture: Qwen2ForCausalLM Vocab Size: 152064 Hidden Size: 3584 Num Hidden Layers: 28 Num Attention Heads: 28 Max Position Embeddings: 32768 Attention Implementation: flash_attention_2 Quantization: 8-bit (load_in_8bit=True)

Tokenizer Configuration

Tokenizer Class: Qwen2TokenizerFast Vocabulary Size: 151643 Model Max Length: 32768 EOS Token: <|im_end|> (ID: 151645) Special Tokens: Check model card for specific template/tokens. Padding Token: <|endoftext|> (ID: 151643)

About Tesslate & Our Vision

Hi everyone, I’m Manav, founder of Tesslate, and we’re on a mission to revolutionize AI by putting powerful reasoning models into your hands.

Today, the AI landscape is dominated by massive frontier models—large, costly, and slow. At Tesslate, we see things differently. The next wave of AI disruption won’t come from sheer size; it'll be driven by speed, specialization, and precision reasoning. Smaller, specialized models aren’t just faster—they’re smarter and more efficient.

Our story began when we released a UI-generation model on Hugging Face that didn't just replicate patterns—it could reason through entire component hierarchies. It resonated instantly, hitting over 10,000 downloads in weeks. That early success validated our vision, and we doubled down.

At Tesslate, we build lean, intelligent models that:

  • 🧠 Think like human agents
  • 💡 Reason through complex, real-world workflows
  • 💻 Execute like elite developers, designers, and analysts

We've already delivered:

  • UIGEN-T1.5: Creating stunning, editable interfaces (React, Tailwind, Three.js)
  • Tessa-T1: A specialized reasoning engine optimized for React development and AI agents (You are here!)
  • Synthia S1: Our flagship general-reasoning model, proving powerful reasoning capabilities beyond STEM into creativity and storytelling.

Our vision is bigger. We aim to do everything covering training, inference, real-time agent actions, infrastructure, research, and innovative products. We’re already piloting with industry-leading clients tackling everything from sophisticated design systems to real-time analytics.

Join us! We're seeking strategic advice, introductions, compute resources, and capital. 👉 Visit tesslate.com to learn more and connect.

Connect with Tesslate
✨ Example Prompts (Click to Load)
System Prompt (Persona & Instructions) 💬 Your Request 🌡️ Temperature 📊 Max New Tokens 🏅 Top-p (nucleus) 🦜 Repetition Penalty 🏆 Top-k 📉 Min-p (Not Active)