Haiku, Sonnet, Opus and Fable are built for different jobs. Pick the wrong one and you burn your usage limit or get a worse answer. Here’s exactly when to use each.
The current Claude lineup is Haiku 5.5, Sonnet 5.5, Opus 5.5 and Fable 5.1. Older models like Opus 5 and Sonnet 4.6 are now legacy. (Anthropic’s model overview)
Why it matters: bigger models use your limit faster. On Anthropic’s API, Opus 5.5 costs twice what Sonnet 5.5 does per token, and Fable 5.1 costs 2.5x Opus. In the app it all comes out of your plan’s usage, so a heavy model on a simple task is wasted usage. (Pricing)
To switch, click the model name next to the send button and pick one. Use More models for the rest. You can switch mid-chat, and it applies from Claude’s next reply.
01
Sonnet 5.5, your everyday workhorse
Anthropic calls it the best mix of speed and intelligence. It’s fast, it’s on every plan including Free, and it handles most daily work really well.
Use it for emails, rewrites, summaries, brainstorming, research synthesis, templates, quick Q&A and building Skills and Projects.
02
Opus 5.5, for the work that matters
Built for long, multi-step work and deeper analysis. Anthropic’s own docs say to start with Opus 5.5 if you’re unsure. It’s on Pro and up, and it’s slower than Sonnet and heavier on usage.
Use it for strategy and big plans, financial, legal or technical analysis, hard writing that needs judgment, diagnosing tricky problems and long documents.
03
Fable 5.1, for the hardest problems
Anthropic’s pick for demanding reasoning and long agent work. It’s the slowest and priciest. It isn’t on Free. On Pro it runs on pay-as-you-go usage credits, and Max and Ultra include it for up to half your weekly limit. (Fable on your plan)
Use it for the problem Opus couldn’t crack, or a high-stakes decision you want stress-tested.
04
Haiku 5.5, for quick, simple jobs
The fastest and lightest model. It’s great for sorting, pulling details out of text, quick rewrites and anything you’d hand an intern.
The old Extended Thinking toggle is gone on the 5.5 models and Fable 5.1. Thinking is always on, and you can’t turn it off. What you control now is effort, meaning how hard Claude thinks before it answers. (Model, effort and thinking settings)
Try This First
Before you jump to a bigger model, raise the effort on the one you’re in. It often gets you there for less.
01
Sonnet 5.5, default effort
Writing and editing, summarizing, brainstorming, research synthesis and day-to-day questions. Speed and lots of quick rounds beat deep reasoning here.
02
Opus 5.5, default or High effort
Complex strategy, deep financial, technical or legal analysis, and long documents where Claude has to hold a lot of detail at once.
03
Opus 5.5 on Extra high, or Fable 5.1
Hard decisions with lots of tradeoffs, root-cause problems, and anything Sonnet already failed at. Escalate the model and the effort together.
04
Haiku 5.5, Low effort
Sorting a list, pulling names and dates out of text, quick reformatting and anything repetitive.
Paste this into Claude, fill in the brackets and it builds your own model rules.
01
Using the biggest model for everything
On routine tasks it’s slower, uses more of your limit and tends to over-explain. Opus on a simple email is like driving a Formula 1 car to get groceries.
02
Leaving effort on Max
High effort on quick tasks eats your limit for nothing. Pick it up for the hard problem, then put it back.
03
Blaming the model when it’s the prompt
A shallow answer usually comes from a shallow brief. Add context and be specific about what you want before you upgrade.
04
Never escalating when you’re stuck
If you’re going in circles on a real decision, 10 minutes on Opus or Fable at high effort is worth the usage.
05
Dragging a long thread into a bigger model
Switching mid-chat works, but the new model re-reads the whole thread on your usage. For a big analysis, start a fresh chat with a clean brief.
Plans and prices change often. Check what your plan includes on Claude’s pricing page, and see my Opus vs Fable cheat sheet for more.