5 min read
Have you ever opened ChatGPT, Claude or Gemini, seen a dropdown full of model names, and quietly picked whichever one was already selected? You are in good company. Most solo owners never touch that menu. But that little choice affects how good your answers are, how fast they arrive, and, increasingly, how much you pay. This beginner guide explains what those names mean, in plain English, and gives you a simple rule for choosing.
First, what is an “AI model”?
An AI model is the engine underneath a chat tool. ChatGPT, Claude and Gemini are the apps you talk to; the models are the trained systems that actually read your message and write the reply. Each company now offers several models at once, much like a car maker sells a city car, a family car and a sports car built on similar technology.
The names change often, which is part of the confusion. Right now, for example, OpenAI’s lineup includes a flagship model called GPT-6 Astra alongside lower-cost siblings called Sol and Luna, according to DataCamp’s coverage of OpenAI’s latest launches. Anthropic’s Claude comes in Opus, Sonnet and Haiku versions. Google’s Gemini has Pro and Flash. You do not need to memorize any of this. You only need to understand the pattern underneath.
The pattern: big, medium and small
Almost every AI company sorts its models into three rough sizes. “Size” here means how much computing power the model uses to think about your request.
- Large (flagship) models are the most capable. They handle complicated reasoning, long documents and tricky writing best. They are slower and cost the most to run.
- Medium models are the everyday workhorses. They are nearly as good for most business tasks and noticeably faster.
- Small (fast) models are built for speed and low cost. They are great at simple, repetitive jobs and can struggle with nuance.
Many tools also offer a “thinking” or “reasoning” mode, where the model takes extra time to work through a problem step by step before answering. Think of it as asking someone to sleep on a decision rather than answer on the spot.
Why this matters more for you now
For a long time, this choice barely mattered to solo owners. You paid a flat monthly fee, and the app picked a model for you. Two things have changed.
First, prices. AI has become dramatically cheaper to run, which we covered in why falling AI costs matter more than any new feature. But the gap between the biggest and smallest models is still large, and you can see it plainly on public price lists like OpenAI’s API pricing page, where small models cost a fraction of flagship ones.
Second, usage limits and metered plans. Many subscriptions now give you a set allowance of the top model, then slow you down or charge more. Using the flagship for every quick question burns through that allowance on tasks that never needed it. Our piece on preparing for metered AI pricing explains this shift in more detail.
A simple rule: match the model to the stakes
Here is the rule to remember: the more it costs you if the answer is wrong, the bigger the model you should use. Here is how that plays out for typical solo business tasks.
Use a small, fast model for
- Fixing grammar and spelling in an email
- Rewording a social post three different ways
- Summarizing a short article or meeting note
- Sorting, tagging or formatting a list
- Quick factual reminders where you will double-check anyway
Use a medium, everyday model for
- Drafting client emails, blog posts and newsletters
- Writing product descriptions or service pages
- Brainstorming offers, names or content ideas
- Turning rough notes into a structured document
Use a large model, or thinking mode, for
- Reviewing a contract or proposal for risks
- Working through pricing decisions or financial scenarios
- Analyzing a long document, spreadsheet or set of reviews
- Planning a complex project with many moving parts
- Anything you would normally pay a professional to check
A useful habit: if a smaller model gives you a weak answer, do not keep rephrasing your prompt five times. Switch up a size and ask once.
Your first step this week
You do not need a new tool to start. Try this short experiment in whatever AI app you already use.
- Find the model menu. In most apps it sits at the top of the chat window or next to the message box. Look at what is available on your plan.
- Pick one real task you do every week, such as writing a client update.
- Run the same prompt on the smallest and largest model you have. Compare the answers side by side.
- Ask yourself honestly: was the bigger answer better enough to justify the extra wait (and, on some plans, the extra cost)?
Many people are surprised to find that, for routine tasks, the fast model is perfectly fine. Others discover their important work has been getting second-rate answers because the default was set to a small model. Either way, you will now be choosing on purpose.
Gentle warnings for beginners
- Bigger is not the same as correct. Even the best model can state wrong facts confidently. For anything legal, financial or medical, treat the answer as a first draft for a professional, not a final word.
- Newer is not always better for you. When a model is replaced, the new one may write in a slightly different style. If something stops working well after an update, it is fine to switch back if your app allows it.
- Do not chase every launch. Model names will keep changing every few months. The big, medium, small pattern will not.
If you are completely new to AI and this still feels like a lot, start with our guide to the first three jobs to hand AI when you run solo, then come back to this when you are ready to fine-tune.
Try the experiment above and tell us in the comments: which task surprised you most, the one where the small model was plenty, or the one where only the big model got it right?
