Claude Opus 5 or Sonnet 5: which to pick per task (July 2026)
Opus 5 shipped July 24 at half the API price of Fable 5. Here is how it differs from Sonnet 5 and which model to pick for real work tasks.
Short answer: pick Sonnet 5 for everyday work, Opus 5 for complex multi-step tasks. Below is why, and where the line runs.
What are these two models?
Both are in Anthropic's current lineup per the official model table: Fable 5, Opus 5, Sonnet 5, Haiku 4.5.
Sonnet 5 is the workhorse: texts, emails, tables, document breakdowns, most daily tasks. Fast, available on every plan.
Opus 5 shipped July 24 and took the top spot for complex work: strategy, long reasoning chains, large documents, code. The previous Opus 4.8 moved to Legacy. In the API Opus 5 is priced at $5 per million input tokens, half of Fable 5.
Where does the line run in practice?
The rule is simple: start with Sonnet 5. If the answer feels shallow, the model loses the thread of a long task, or oversimplifies a complex document, switch to Opus 5 and rerun the same prompt. The difference shows immediately on multi-step work: a contract with cross-referencing clauses, a quarter strategy under constraints, debugging a long process.
Since the release, the community has been actively comparing the two at high effort levels. The practical takeaway doesn't change: test on your own task, switching takes two clicks in the selector at the bottom of the input box.
| Task | Model |
|---|---|
| Emails, posts, tables, quick breakdowns | Sonnet 5 |
| Strategy, large documents, code, automations | Opus 5 |
| Very short and instant | Haiku 4.5 |
| Extremely hard cases Opus 5 couldn't crack | Fable 5 (credits on Pro now) |
What about effort levels?
Models have a depth-of-thinking dial. Higher effort: deeper thinking, slower answers. Medium is enough for quick tasks, go higher for complex ones. More on the Opus 5 release and the rest of July's updates in the full breakdown.
We train picking the model per task hands-on in our free guides and the first lesson with no signup.
AGINE Academy is an independent product, not affiliated with Anthropic. Claude is a product of Anthropic.
Questions
No. It goes deeper on complex multi-step tasks, but for everyday work Sonnet 5 is faster and more than enough. Working rule: start with Sonnet 5, switch to Opus 5 when an answer feels shallow or the task is long.
Yes, both Sonnet 5 and Opus 5 are available on paid plans on the web, in the app and via the API. Pick them in the model selector at the bottom of the input box. Only Fable 5 runs on separate credits on Pro.
Per Anthropic's official model table: $5 per million input tokens, half of Fable 5 at $10. If you work through the app, pricing doesn't matter, it's all included in the subscription.