What are parameters in an AI model?
Quick answer
Parameters are the numbers inside an AI model that get adjusted during training, sometimes called weights. Together they hold everything the model learned. A model "with 175 billion parameters" has that many learned numbers. More parameters often means a more capable model, but also a slower and more expensive one to run.
Last updated
Take the free module →Module 2 of the free Claude Jump Start course, about 7 minutes. No account needed.
Updated · By Robert Breen
Why it matters for a small business
Parameter counts used to be the headline number for a model. OpenAI's GPT-3, published in 2020, had 175 billion. Today many companies, including OpenAI, Anthropic and Google, don't publish counts for their newest models, so you usually compare models by what they are for, their price and how they do on your own task.
The practical trade-off still holds. Large models handle harder reasoning and long documents better; small models answer faster and cost less per token. For routine work like a short customer reply, a smaller model is often enough.
In a real lesson: Your First Chat with Claude: Models, Files and a Great First Prompt
In Your First Chat with Claude, the model picker describes each model by its job rather than its size: Fable for the toughest challenges, Opus for complex and everyday work, Sonnet as the most efficient for simpler tasks, and Haiku as the fastest for quick answers. Anthropic doesn't list parameter counts.
The lesson's advice works without them: for customer emails and most small business writing, Opus or Sonnet is a good choice, and you can switch models at any point in a chat. That is the habit to keep. Match the model to the job, then check the answer.
Try this lesson free or read the step-by-step guide.
Common confusions
Model parameters vs settings
Parameters are learned during training and you can't change them. Settings you control, like temperature or maximum tokens, are sometimes also called parameters in APIs, which causes the mix-up.
Parameters vs tokens
Parameters are inside the model. Tokens are the pieces of text going in and out, and they are what you are billed for.
Tips
- Start with the smaller, cheaper model and move up only if the answers fall short.
- Compare models on three of your real tasks rather than on size claims.
- In an API or n8n node, "parameters" may mean request settings, not model size.
Related terms
More AI basics terms
Where you use it: free lessons
- Your First Chat with Claude: Models, Files and a Great First Prompt (Claude, 7 min)
- Claude Projects for Your Business: Instructions, Knowledge and Quotes (Claude, 8 min)
Frequently asked questions
- Does a bigger model always give better answers?
- Not always. Bigger models are usually stronger on hard tasks, but a well-prompted smaller model can match them on simple, routine work.
- How many parameters does ChatGPT have?
- OpenAI hasn't published parameter counts for its current models. The last widely cited official figure is GPT-3's 175 billion, from 2020.