Anthropic ships new Claude models every few months, and most people never change the drop-down. Here is what each one is actually for.
If you have opened Claude lately and seen a model picker with names like Haiku, Sonnet, and Opus. You are not alone in ignoring it. Most people pick whatever loads by default and never think about it again. That is usually fine, but knowing what each tier is built for can save you real time, and sometimes real money, if your business is paying for API access instead of a flat subscription.
Here is the current lineup as of August 2026, and when each one actually makes sense.
The Lineup
- Claude Haiku 4.5 – the fast, cheap option. Built for quick, simple tasks where speed matters more than depth.
- Claude Sonnet 5 – the new everyday default. Launched June 30, 2026, it is now the default model on Free and Pro plans and handles most day-to-day work well.
- Claude Opus 5 – the step up for harder work. Launched July 24, 2026, it is now the default on Claude Max and the strongest model available on Pro, and independent coverage confirms it is priced the same as its predecessor but noticeably more capable.
- Claude Fable 5 – the top tier. The most capable model Anthropic offers generally, at a higher price point, for the small slice of work that genuinely needs it.
When to Use Each One
Haiku 4.5 is the right call for anything high-volume and low-stakes: sorting a batch of emails, a quick first pass at a document, drafting a short reply, or any task you would not think twice about if a fast, competent junior handled it. If you are doing the same simple task dozens of times a day, Haiku is where you save the most money without losing much quality.
Sonnet 5 is the model most people should default to for real work. Writing a client email, drafting a report, reviewing a contract’s first pass, building a spreadsheet formula, most day-to-day business tasks land here. It is close to Opus-level performance at a lower price, and it is agentic enough to handle multi-step tasks without hand-holding.
Opus 5 earns its keep on the harder, messier problems: a complex financial analysis, a long research task, debugging a tricky process, or anything where you have tried Sonnet and it fell short. Anthropic built Opus 5 specifically to bring near-top-tier intelligence to everyday coding and knowledge work at a price that no longer makes you think twice about using it.
Fable 5 is worth reaching for only when the task genuinely demands maximum capability: the hardest analytical work, long autonomous research, or a problem where you have already tried Opus and need more. For most small business tasks, you will not need it.
Not sure which model fits a task you are working on? That is exactly the kind of question we help clients sort out as part of setting up their AI tools the right way.
Don’t Overlook the Effort Setting
Right next to the model picker in Claude.ai sits a second, easy-to-miss control called Effort. It decides how much time Claude spends thinking before it answers, and it matters almost as much as which model you picked.
| Effort Level | What It Does | Best For |
|---|---|---|
| Low | Fastest, most direct answers | Quick chats, simple questions, high-volume tasks |
| Medium | A cost-conscious step down from the default | Routine drafting and everyday requests |
| High (default) | Balanced reasoning for most real work | Reports, analysis, multi-step tasks |
| Extra | Deeper reasoning before responding | Complex, high-stakes work worth the extra time |
| Max | Maximum reasoning, no shortcuts | The hardest problems, where being right matters more than being fast |
A quick way to think about it: the model you pick decides how capable Claude can be, and the effort level decides how hard it actually tries on this particular question. A high-effort Sonnet 5 can outperform a low-effort Opus 5 on the same task.
A Simple Rule of Thumb
Start with Sonnet 5 at the default High effort for anything that is not trivial. If the output feels shallow, bump effort up to Extra. Before you switch models entirely, this is often the faster fix. If a multi-step task still struggles after that, move up to Opus 5. Save Haiku 4.5 and Low effort for repetitive, simple work you do at volume, and only reach for Fable 5 or Max effort when a task has already beaten your other options.
What This Does Not Mean
Model choice is not a substitute for a good prompt. A vague request to a top-tier model often does worse than a clear, specific request to a mid-tier one. Before assuming you need a more expensive model, try being more specific about what you actually want first.
Where GiaSpace Fits
We help small to medium sized businesses figure out which AI tools and models actually fit their workflow, instead of guessing or overpaying. If you want a second opinion on your AI setup, that is where we come in.
→ Schedule a Free Assessment with Rob Giannini, our CEO and AI specialist
