Open Claude for the first time and most people stall at the very first step: choosing a model. Four names sit there — Fable, Opus, Sonnet, Haiku — and nothing says which one fits the job you have. Pick at random and you worry about wasting money; sit down to read the English documentation and the morning is gone.
This article explains, briefly, what Claude is, how the four models differ, and most importantly how to use each one. When you finish it, you will know where to start.
Claude is a line of AI models made by the company Anthropic, and it has four models. Each model is a separate AI: you type a message in, and it writes an answer back. Put simply, Claude is not a single chatbot but four different AIs, each strong in its own way. The difference between the four models comes down to three things: how deeply they reason, how fast they answer, and what kind of work they fit.
Anthropic currently has four models, arranged in four tiers. The higher the tier, the stronger the model but the slower its answers; the lower the tier, the faster and cheaper it is. The information in this table was verified in November 2026.
| Model | What it is for | Speed | Token price (USD / 1 million tokens, in / out) |
|---|---|---|---|
| Fable 5.2 | Hard work that needs deep reasoning and runs for hours | Moderate | $8 / $40 |
| Opus 5.5 | Writing code and working with knowledge over long periods | Moderate | $4 / $20 |
| Sonnet 5.5 | Everyday work that needs to be both fast and good | Fast | $2 / $10 |
| Haiku 5.5 | Extremely simple work, repeated at large volume, such as classifying and extracting | Fastest | $0.10 / $0.50* |
These are API prices, in USD per 1 million tokens. Haiku: for prompts over 100,000 tokens, the price rises to $0.50 in / $2.50 out.
Fable sits at the top tier. It is Anthropic’s strongest model, built for work that needs deep reasoning or runs on its own for hours without you sitting next to it — but it is also the model that burns through tokens fastest, so do not pick it for everyday work. Opus sits just below Fable and is the default choice Anthropic recommends. Sonnet is the everyday model: fast enough to feel almost instant, and strong enough for most writing and coding. Haiku is the smallest, fastest to answer and cheapest, and it fits extremely simple work repeated thousands of times a day.
One small note: older models such as Fable 5.1 and Opus 4.8 are still there for existing users, but Anthropic has put them in the legacy group. For new work, start with the four models in the table above.
Besides the four models above there is also Mythos 5.2, which Anthropic launched at the same time as Fable 5.2 on 9 November 2026. It is said to be up to 30% faster than version 5.1 and 20% cheaper, but it is only for vetted organisations in cybersecurity or biomedicine, so ordinary users will not see it on claude.ai.
Many new users tend to pick the newest model, assuming newest means best. Anthropic advises users to start with Opus 5.5, but my own experience is a little different.
In my view, Sonnet and Opus are the two models most worth using. Start with Sonnet first, though: it is fast, and its quality is good enough for most everyday work. Only move up to Opus when Sonnet does not handle a request well. If Opus still does not meet your needs, that is when to consider Fable, because this model burns through tokens very fast.
Put simply, the rule is this: the clearer and more repetitive the job, the lower down the tiers you should go; only move up when the job is hard enough. The three most common cases:
Pick the right door for the job you are doing. There are three:
claude.ai is the most familiar door. Open your browser, go to claude.ai, sign in and type your question into the chat box. The model you are using shows right on the chat box, and you click it to change models. This is the best place to chat, ask questions, draft text and try out an idea quickly.
Claude Code is Anthropic’s coding tool. Unlike the web chat, it works directly on the files in your project, so it fits writing code, fixing bugs and making changes spread across many files. This is the place for people who write code, not for general chatting.
API is the door for developers. Instead of opening the website to chat, you write a program that calls Claude directly to put it inside your own application — for example, a customer-support chatbot or a tool that summarises documents automatically. API stands for application programming interface: a set of rules that lets one program ask another program to do work.
If you just want to try Claude for the first time, open claude.ai. You do not need to touch the other two yet.
Yes. You can chat with Claude for free on claude.ai. The free plan has usage limits, and the paid plans raise those limits.
No. Fable 5.2 is the newest and strongest, but for everyday work Sonnet 5.5 answers faster, and for extremely simple work at large volume Haiku 5.5 fits better. A newer model helps when the work is harder, not in every case.
Start with Sonnet 5.5. For work that is extremely simple and repeats itself, drop down to Haiku. For work so hard that Sonnet does not do it well, move up to Opus, and only try Fable when Opus still does not handle it — and remember that Fable burns through tokens very fast. Open claude.ai, pick Sonnet 5.5 and get started. You only need to think about changing models once you have hit a job Sonnet does not do well.
Verified 10 November 2026; Mythos and token prices verified 12 November 2026.
Full-stack developer
You hand an AI a programming task that takes many steps. At first, it works fast and seems to understand the request. But the longer it goes, the more the result drifts: the original requirements gradually get dropped, code that was never broken gets changed, and the AI even reports that tests passed when the...
You might be using Chrome to browse the web, Edge for work, or VS Code to write code without realising they share a common foundation: Chromium. It is the open-source browser project behind many familiar products — but Chromium and Google Chrome are not quite the same thing. This article explains what Chromium provides, what...
The 11.11 sale season is peak time for online shopping scams. Learn to spot 8 common tricks, including AI deepfakes, and how to protect yourself as a buyer or a shop owner.
Comments
You can use your Google account to comment. Google only shares your verified name and email address to display your name. You can log out at any time.
Sign in with GoogleNo comments yet. Be the first!