You may wonder if there’s a possibility to create your own AI chatbot legally and without paying OpenAI or Anthropic or other company a dime. The answer is yes. Moreover, you can do it without installing anything and by just using your e-mail account. In this article we explain you how to.
The key element for creating your own AI chatbot without paying API from Artificial Intelligence developers are the GGUF models. These are optimized versions of the great languge models that you can run directly from your PC. Even if your hardware is not powerful enough, you can still run it thanks to the optimization levels.
The Nebula Block tool for AI chatbot creation
Nebula Block is a very simple tool to help you create an AI chatbot. A complete platform that gives you access to endpoints which allow sending messages and receive answers provided by the language model. You can also rent a GPU whenever your free credits run out.
In order to create your own AI chatbot go to the Nebula Block website. It’s available in English and French but the AI model supports many other languages. Once you are online, select the model you want to run. There is a wide variety, from multimodal AI like Claude Sonnet 4, Gemini-2.5 Pro and GPT 4o-mini to text models like DeepSeek-R1 or Llama.3. On the right screen you can see how much does it cost to use them. If you selet DeepSeek-R1-0528 for example, you will have for free.
Setting your own AI chatbot easily
From the setting screen you will access a main panel with a box to input your e-mail account. This is the only requirement you need to complete in order to use the model. From the right panel you can access a list of options:
Model
Choose between the different AI models available on the web. In the little box below you can see which language you are using right now.
System Prompt
Give instructions to the AI and contextualize the answers. You can indicate if you want a serious, joking or scientific tone for the answers, or even if you prefer a more familiar talking pattern.
Adjustment bars
The first one allows you to choose the token extension for each answer. In Temperature you can tweak the answer’s creativity. The more temperature, more random examples in the answers. In Top P you can select more repetitive structures for each text or unique ones. The higher Top P bar is, the more diversity you will find in your text structure. A good tip, don’t move up Temperatura and Top P at the same time.
The Nebula Block free options
All the features you see in Nebula Block for free are more or less what you read above. But there’s also the possibility of accesing a paid model. You can rent it at 40 cents the hour depending on the language model you select. The most common conditions for paid models are queries limit, more or less 20 a day. If you exceed the free credits from the website, you must pay the rent.
The most popular models and what they do
- DeepSeek-R1. It’s free with certain limits and you can use it for code and technical thinking.
- Llama 3-3. A generalist language model with conversationalist options.
- Qwen-QwQ. Excellent for translation and bilingual content.
- Gemini-2.5 Pro. Multimodal tasks and complex analysis.
Do you have any questions or problems related to the topic of the article? We want to help you.
Leave a comment with your problem or question. We read and respond to all comments, although sometimes it may take a while due to the volume we receive. Additionally, if your question inspires the writing of an article, we will notify you by email when we publish it.
*We moderate comments to avoid spam.
Thank you for enrich our community with your participation!