🧠 About This AI Chat UI

⚙️ Model Capabilities & Design
• Powered by DeepSeek Coder V2:16b (15.7B parameters)
• Includes DeepSeek V2, CodeLlama by Meta, OpenChat, Deep Cogito, Mistral Nemo, Falcon 3 and Mathstral (Q4_0 and Q4_k_m quantizations)
• Enhanced and accelerated using Ollama

Designed for:
🧠 Code generation and debugging
🔬 Math, science, and technical reasoning
🛠️ Project creation and structured tasks
• Not optimized for casual human-like conversation
• For best results: use clear, direct, and detailed prompts

🧩 Model Selection & Sessions
• Choose your model from the dropdown above the input box
❗ You cannot switch models mid-session — this preserves chat history integrity

To use a different model:
➕ Click Start New Chat
• Select the desired model at the start
• Session history retains the last-used model, even after logout

⚡ Performance & Comparison
• Benchmarked against ChatGPT 4o:
🚀 Comparable or faster response speeds
🧩 Equal or better reasoning in many domains
• Responses use up to 3000 tokens for detailed output
• Contextual memory of your last 2 messages improves continuity

🖥️ User Interface Advantages
• Built for developers, businesses, and power users
• Offers UI features beyond ChatGPT 4o, See below including:
🔎 Session management sidebar
📥 Code downloading
🎨 Image generation
🌗 Dark/light mode toggle

• Intended as a high-quality testbed for future hardware and model upgrades

Features

• 💬 Interactive AI chat interface with ChatGPT-like features and unique enhancements
• 📄 Change the UI language to chinese, japanese, spanish, french, arabic and hindi
• 🎙️ Speech to text for image generation and message prompt text fields
• 🎙️ Voice commands for many functions. See Voice Help Menu
• 🧠 Streaming, auto-scrolling assistant responses with formatted code
• 🕒 Time-stamped user messages and assistant replies
• 🗃️ Persistent chat history per user — sortable by date or renamed title
• 🛠️ Edit and resend user prompts inline — fix or improve previous questions
• 🔁 "Delete Last" button removes your previous message, assistant text, and code in one click
• 📝 Format code snippets using a checkbox for cleaner submission and better parsing
• 📝 Generate images from text:Optional. Requires your own API key but we have it here for you to try
• 📦 Download assistant text as PDF, Markdown, or Plain Text with filename of your choice
• 💾 Download code blocks with correct extension and custom filename
• 📁 Export full chat session as Markdown or PDF
• 📌 Rename chat sessions for better organization (default is timestamped)
• 🆕 Create up to 20 unique project chat sessions per user
• 🆕 Switch between 8 AI models of your choice
• 🚮 Manage Sessions: Clear all sessions, start a new chat, or delete individual ones
• 🧮 Built-in tokenizer: shows token + character count for each response and last 2 totals combined (helps stay under 3000 limit)
• 📋 Copy code with smart cleanup — automatically strips language label and backticks
• 🔐 Secure user login system with roles (admin, guest, etc.)
• 🔐 Global user session tracking for roles (admin, guest, etc.)Login/Logout/Browser Close
• ⏳ Session security: automatic logout after 10 minutes of inactivity or 90 minutes of continuous use
• 💡 Tip: When copying and resending AI responses as new prompts, use the Copy and Format as Code tools for better results

How It Works

Utilizing Flask for backend, HTML and JavaScript for frontend, and Ollama for AI. Chat histories are stored on our server for continuity. Access is global via Cloudflare URLs, with 5 private guest accounts available for testing. Our chat UI can also be run locally on your own PC if equipped with proper hardware and model sizes to match.

Server maintenance, network issues or software development may cause intermittent downtime, so please retry later if issues occur.

Contact Us

For questions,feedback, or to purchase our chat UI source code for personal, business or commercial use, reach out to us at [email protected].