Claude Sonnet vs Opus
Claude Sonnet 5.5 vs Claude Opus 5.5: price, speed, context and which one Anthropic suggests for which job.
Opus 5.5 is built for long-running agentic coding and knowledge work. Sonnet 5.5 is Anthropic's best combination of speed and intelligence. Sonnet costs half as much per token ($2 and $10 per million input and output tokens, against $4 and $20 for Opus) and Anthropic rates it faster. Both take up to 1 million tokens of input (as of November 2026).
Our advice: use Opus 5.5 for long, hard agent work such as multi-hour coding or large refactors, and Sonnet 5.5 for everyday coding and tool use when speed and cost matter. Anthropic's guide says most workloads can start on Opus 5.5 and move to Sonnet where your tests show it is good enough. For how the Claude plans compare with ChatGPT's, see Claude vs ChatGPT.
Good for
- Developers picking a model for the Claude API
- Claude Code users choosing between Opus and Sonnet
- Claude app users deciding if Pro is worth it for Opus
- Teams watching their token bill
Side by side
Opus 5.5 costs exactly twice Sonnet 5.5 per token. API prices as of November 2026 (checked November 3, 2026).
| Option | Sonnet 5.5 | Opus 5.5 |
|---|---|---|
| API model ID | claude-sonnet-5-5 | claude-opus-5-5 |
| Input, per 1M tokens | $2 | $4 |
| Output, per 1M tokens | $10 | $20 |
| Speed | Fast | Moderate |
| Context window | 1M tokens | 1M tokens |
| Default effort (API) | high | medium |
| Fast mode | Not listed (Opus models only) | Yes, $8 input / $40 output |
| Claude app plans | Free and up | Pro and up |
Which to pick
Our advice: start on Opus 5.5 for agentic coding and complex work, as Anthropic's guide suggests, then move steps that pass your tests to Sonnet 5.5 to halve the cost. For chat and quick edits, Sonnet 5.5 is fast and is on the Free plan.
Before switching models, try changing the effort setting: Anthropic says that is often the better lever.
Best for
- Opus 5.5: multi-hour coding agents and large refactors
- Opus 5.5: complex systems work, vision-heavy tasks and computer use
- Sonnet 5.5: everyday code generation and data analysis
- Sonnet 5.5: agent tool use where speed and cost matter
Skip it if
- Opus 5.5: you are on the Claude Free plan (Opus starts with Pro)
- Sonnet 5.5: you want API fast mode, which lists Opus models only
Questions
Is Opus better than Sonnet?
Anthropic builds Opus 5.5 for long-running agentic coding and knowledge work, and describes Sonnet 5.5 as its best combination of speed and intelligence. Opus costs twice as much per token. Our advice: test both on your own prompts before paying for the larger model.
How much more does Opus cost than Sonnet?
Twice as much. On the API, Opus 5.5 is $4 per million input tokens and $20 per million output tokens; Sonnet 5.5 is $2 and $10 (as of November 2026). Cache reads are $0.20 and $0.10, and the Batch API halves both prices.
Which model does Claude Code use, Opus or Sonnet?
Opus 5.5 by default on Pro, Max, Ultra, Team, Enterprise and the API. Type /model sonnet to switch, or /model opusplan to plan with Opus and write the code with Sonnet.
Can I use Opus on the free Claude plan?
No. The Free plan lists Sonnet and Haiku; Opus starts with Pro at $20 a month, or $17 a month billed yearly (as of November 2026). For how the plans compare with ChatGPT's, see Claude vs ChatGPT.
Is Sonnet faster than Opus?
Yes. Anthropic rates Sonnet 5.5 as fast and Opus 5.5 as moderate. On the API, Opus 5.5 has a fast mode with up to 2.5 times the output speed, priced at $8 input and $40 output per million tokens (as of November 2026).
Do Opus and Sonnet have the same context window?
Yes. Both take up to 1 million tokens of input at the standard price, and both write up to 128,000 tokens per reply.
Where do Haiku and Fable fit?
Haiku 5.5 is the fastest and cheapest, from $0.10 input and $0.50 output per million tokens, for high-volume simple work. Fable 5.1 sits above Opus at $10 and $50, for the hardest long-running work (as of November 2026). In the Claude app, Pro pays for Fable with usage credits and Max includes it up to half the weekly limit.
What is effort, and should I change it instead?
Effort sets how much the model thinks before answering, trading quality for speed and cost. On the API, Sonnet 5.5 defaults to high and Opus 5.5 to medium; Claude Code starts both at medium. Anthropic says tuning effort is often a better lever than switching models.
Sources
Docs, issue trackers and discussions this page is built from.
Show all 5Hide
- Models overview · platform.claude.com
- Choosing a Claude model · platform.claude.com
- Claude API pricing · platform.claude.com
- Claude plans and pricing · claude.com
- Claude Code model configuration · code.claude.com