Overview
An AI model is a system trained on large amounts of data to recognize patterns and generate output — text, images, code, audio, or video — based on an input prompt. "Model" here refers to the underlying trained system, distinct from the product or app (like a chat interface) built around it.
Main Types of Models
Most models fall into a handful of broad categories: large language models (LLMs) for text generation and reasoning, reasoning models tuned specifically for multi-step logical or mathematical problems, image models for generating or editing visuals, video models for generating short clips, audio models for speech and music, code models tuned for programming tasks, and multimodal models that handle several of these input or output types at once.
How to Think About Choosing One
The right model depends far more on your specific task than on a general leaderboard ranking. A model that tops a broad benchmark may still be a poor fit if your task needs a specific strength — long context, low latency, a specific language, or a particular cost profile — that the benchmark doesn't measure. See our Models directory for individual profiles, or Comparisons to weigh two specific options directly.
Open-Weight vs. Closed Models
Some models are released as "open-weight," meaning the trained parameters are publicly downloadable and can be run on your own infrastructure; others are "closed," accessible only through the provider's own API or app. Open-weight models offer more control and no per-token API cost once hosted, but require you to manage your own infrastructure; closed models are simpler to start with but tie you to the provider's pricing and availability.
Related Pages
Frequently Asked
What's the difference between a model and an app like ChatGPT?
The model is the underlying trained system; ChatGPT is a specific product built around one (in this case, OpenAI's GPT models), with its own interface and features layered on top.
How do I know which model is best for my use case?
Match the model's specific strengths to your actual task rather than a general leaderboard rank; see our Comparisons hub for direct, task-relevant trade-offs.
What does 'context window' mean?
It's the amount of text (measured in tokens) a model can consider at once in a single conversation or request; a larger context window lets it handle longer documents or conversations without losing earlier information.
Where can I see specific models tracked by LLMWIKI?
See the Models directory for 71+ individual model profiles across LLM, reasoning, image, video, audio, code, multimodal, and embedding categories.