Artificial Intelligence has become more accessible than ever thanks to Together.ai, which provides free access to powerful models like Llama 3.3 70B Instruct Turbo. In this publication, I’ll walk you through how I built a standalone chatbot application that runs on Windows as a packaged .exe — no Python installation required!

This project demonstrates how to:
✅ Connect to Together.ai API and use cutting-edge models.
✅ Stream chatbot responses in real-time with a typing effect.
✅ Build a command-line chatbot interface with personality and style.
✅ Package the chatbot into a portable executable for easy sharing.

Introduction

As AI developers, one of the challenges we face is accessibility. Running large models locally requires expensive GPUs, while API access often comes with strict paywalls.

Together.ai bridges this gap by offering free access to the Llama 3.3 70B Instruct Turbo model — one of the most powerful open-source LLMs currently available.

This project started as an experiment to see if I could:

• Build a lightweight chatbot powered by a cutting-edge model.
• Make it accessible to non-developers by packaging it into a .exe.
• Ensure it works offline once built (only requiring internet for API calls).

The result is an AI chatbot you can run on any Windows machine.

Why Llama 3.3 70B?

The Llama 3.3 70B Instruct Turbo model is:

• Massive (70B parameters): Offers high-quality reasoning, coherence, and long-context handling.
• Optimized ("Turbo"): Runs faster thanks to FP8 quantization while supporting up to 131k tokens of context.
• Free on Together.ai: Developers can explore without upfront cost, though with some rate limits (6 requests/minute).

This makes it an excellent choice for chatbots, assistants, and research projects.

Project Features

• Real-time Streaming → The chatbot “types” responses as they are generated.
• Emoji & Unicode Support → Fun, modern responses with personality.
• Cross-Version Compatibility → Works on Windows 7 and above.
• Packaged Executable → End-users don’t need Python installed.
• ASCII Art Branding → Custom startup splash with your bot’s name/logo.

Installation & Setup

Clone the repository:
git clone https://github.com/affan4321/AI-Chatbot-using-together.ai-API-with-llama3.3-70b-instruct-turbo-free.git
cd AI-Chatbot-using-together.ai-API-with-llama3.3-70b-instruct-turbo-free

Install dependencies:
pip install -r requirements.txt

Set your Together.ai API key in a .env file:
TOGETHER_API_KEY=your_api_key_here

Run the chatbot:
python chatbot.py

How the Code Works

At the core of the project is chatbot.py, which:

• Authenticates with Together.ai using your API key.
• Sends user input to the Llama 3.3 70B Instruct Turbo model.
• Streams back tokens in real-time for a typing effect.

Handles Unicode/emoji gracefully, ensuring compatibility across Windows versions.

Building the Executable

To package the chatbot into a .exe (Windows standalone app), run:
python build_executable.py

This uses PyInstaller to generate a distributable file inside the dist/ folder.

Now you can share your chatbot with any Windows user — no Python installation required! 🎉

Security Considerations

• The API key is bundled in the executable. This is fine for personal use but not secure for production.
• For serious deployments, consider:
  -> Running requests via a server-side proxy.
  -> Rotating and monitoring API key usage.
• Windows may flag the .exe as “unverified” because it’s unsigned. This is normal for independent developers.

Limitations of the Free Tier

• Rate Limits: 6 requests/minute (free users), 10/min (paid users).
• No Custom Fine-Tuning: You can’t train your own model; only use what Together.ai provides.
• Internet Dependency: The chatbot requires an internet connection to access the API.

Future Work

Some directions I plan to explore:

• 🌐 Web-based UI (React + FastAPI backend).
• 📱 Mobile version with Flutter.
• 🔑 Secure API proxy layer for multi-user deployments.
• 🤖 Plug-and-Play Personalities → Different system prompts for various chatbot roles.

Conclusion

This project shows how any developer can leverage state-of-the-art AI models for free and package them into a form that non-technical users can run instantly.

By combining Together.ai’s Llama 3.3 70B Instruct Turbo with Python, streaming APIs, and PyInstaller, I created a portable AI chatbot that works out of the box.

💡 This is just the beginning. With the right tools, we can bring advanced AI closer to everyone — not just developers with powerful GPUs.

🔗 GitHub Repo: AI Chatbot using Together.ai API with Llama 3.3 70B