Written by Martin Ågren - April 13, 2026
An AI model (ML model) is trained via machine learning to learn patterns and relationships from training data in order to produce relevant output (for example classification or something generated) without being explicitly programmed for every possible input.
Regarding AI chatbots (such as ChatGPT):
Inside an AI chatbot, there can be several AI models. Imagine having different AI “personalities” inside an AI chatbot that you can give instructions to and chat with.
Different AI models can, for example, be specialized in different things. One AI model can be an expert at generating text, while another AI model can be an expert at generating images, etc.
Other AI models can have a certain behavior that is sometimes beneficial to use, for example reasoning models.
Singular: LLM
Plural: LLMs
A type of AI model that is often talked about nowadays is LLMs. LLMs are specialized in language. LLMs can both understand and generate language.
An LLM has been trained on large amounts of text to learn how both natural (human) language and coding languages are structured.
Weights in neural networks in LLMs:
Through training, an LLM develops representations of relationships between words and their meanings via "weights" (more or less strong connections between neurons in an artificial neural network). It can then generate tokens (parts of words) using probability calculations.