What is a model family?
Quick answer
A model family is a group of related AI models released by one company under one name, usually in several sizes. The big model is the most capable and the most expensive; the small ones are faster and cheaper. GPT, Claude, Gemini and Llama are model families.
Last updated
Updated · By Robert Breen
Why it matters for a small business
Families explain the confusing names. Within Claude, as of October 2026, Anthropic's docs list Fable and Opus at the top, Sonnet in the middle and Haiku as the small, fast tier. Google's Gemini family has Pro, Flash and Flash-Lite. OpenAI's GPT models come as a flagship plus "mini" and smaller variants. Meta's Llama family ships in sizes measured by their number of parameters.
Knowing the family lets you trade cost for ability without changing tools. In a chat app, the model menu is the family. In an automation, it is the model dropdown, and switching from a large to a small model in the same family usually keeps the same prompts working. Through an API, the smaller tiers can cost a fraction of the top tier per token, which adds up when a workflow runs hundreds of times a day. Version numbers move several times a year, so a family name is a better guide than a specific model number in a tutorial.
Three families and their tiers, as of October 2026
| Family | Most capable | Everyday | Fast and cheap |
|---|---|---|---|
| Claude (Anthropic) | Fable, Opus | Sonnet | Haiku |
| Gemini (Google) | Pro | Flash | Flash-Lite |
| GPT (OpenAI) | Flagship GPT model | Standard GPT model | Mini models |
In a real lesson: Your First Chat with Claude: Models, Files and a Great First Prompt
Step two of the Your first chat with Claude lesson is called Look at the models. It reads the Claude family from top to bottom: the top models "for the hardest, most complex work," Sonnet as the everyday model, and Haiku as the fastest. Then it adds a warning worth remembering for every family: names change with new versions, and the Free plan shows fewer.
Try this lesson free or read the step-by-step guide.
Common confusions
Model family vs model version
The family is the name, such as Claude or Gemini. The version is the number, such as 5.5 or 3.1, and it changes with each release.
Model family vs product
Gemini is both a model family and Google's app. The app can switch between family members for you.
Tips
- Test the middle tier first; it is the best value for most office work.
- When a vendor releases a new version, retest your saved prompts.
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)
- n8n AI Agent Tutorial: Save Social Media Ideas to Google Sheets (n8n, 12 min)
Frequently asked questions
- Why do AI companies release several sizes of the same model?
- Because jobs differ. A big model handles hard reasoning but costs more and runs slower. A small one answers simple questions almost instantly for much less. Offering a family lets each customer pick the trade-off.
- Can I mix model families in one workflow?
- Yes. Tools like n8n let each step use a different model, even from different companies. Many teams use a cheap model to sort and a stronger one to write.
- Do prompts work the same across a family?
- Mostly. A clear prompt usually works on every tier of a family, but smaller models follow long or complicated instructions less reliably, so test before you switch.