Claude Sonnet 5.5 vs Claude Sonnet 5
Which one to choose
Claude Sonnet 5.5 is Anthropic's successor to Claude Sonnet 5, so pick Sonnet 5.5 for new work and for any agent you want faster and cheaper per task. Pick Sonnet 5 only to hold behaviour steady in an integration already tuned to it. Sonnet 5.5 is the current Sonnet, with faster output and a large gain on agentic coding.
Pricing comparison
| Claude Sonnet 5.5 | Claude Sonnet 5 | |
|---|---|---|
| Model maker | Anthropic | Anthropic |
| Delivery availability | Available | Available |
| Context window | 1M | 1M |
| Max output | 128K | 128K |
| Official price | Input$2.00per 1M tokensOutput$10.00per 1M tokens | Input$3.00per 1M tokensOutput$15.00per 1M tokens |
| TokenLab price | Input$0.60per 1M tokensOutput$3.00per 1M tokens | Input$0.90per 1M tokensOutput$4.50per 1M tokens |
| Model performance |
|
|
| Capabilities | JSON modePrompt CacheReasoningTool useVision | JSON modePrompt CacheReasoningTool useVision |
Choose Claude Sonnet 5.5 when
- You build coding or computer-use agents and want tasks to finish in fewer steps.
- You want faster output on the same tier.
- You want a knowledge cutoff of June 2026.
Choose Claude Sonnet 5 when
- You have evaluations tuned to Sonnet 5 and want identical behaviour.
- You rely on its habit of checking its own output and writing a reproducing test before a fix.
- You compare behaviour across Sonnet generations and need the earlier one as a baseline.
How they differ
| Aspect | Claude Sonnet 5.5 | Claude Sonnet 5 |
|---|---|---|
| Role | The current Sonnet, released September 2026. | The June 2026 Sonnet, replaced by Sonnet 5.5. |
| Agentic coding | Anthropic reports a very large jump on Terminal-Bench 4.0 over Sonnet 5. | Finishes complex tasks where earlier models stopped short, but trails Sonnet 5.5 on Terminal-Bench 4.0. |
| Tool calling | Batches tool calls better, so tasks take fewer steps. | Checks its own work without being asked, for example by writing a test that reproduces a bug first. |
| Safeguards | New cybersecurity safeguards restrict some tasks. | The fallback model when Sonnet 5.5 declines a higher-risk request. |
| Request rules | Rejects non-default temperature, top_p and top_k values and errors on forced tool use. | Also rejects non-default temperature, top_p and top_k values. |
Summary
- Claude Sonnet 5.5: Input $0.60 / Output $3.00 per 1M tokens; Claude Sonnet 5: Input $0.90 / Output $4.50 per 1M tokens. Rates depend on the billing unit, specification, and usage. Compare matching conditions in the model's detailed pricing; a single rate does not determine the total cost.
FAQ
Is Sonnet 5.5 better than Sonnet 5?
Yes. Sonnet 5.5 is Anthropic's successor, with faster output, fewer tokens per task and much stronger agentic coding results than Sonnet 5. New builds should start there.
Do I need to change prompts when moving to Sonnet 5.5?
Task prompts carry over. Check tool-choice settings, since forced tool use returns an error on Sonnet 5.5, and rerun your evaluations before switching traffic. Expect to adjust little else.
Why would anyone keep Sonnet 5?
Stability. If your pipeline is validated on Sonnet 5, staying put avoids retesting. Sonnet 5 also acts as the fallback for some requests Sonnet 5.5 declines.
Do both accept images?
Yes. Both take text and images, call tools and answer in text, so image-based agent steps work on either. That keeps screenshot review in either one.
Which is cheaper, Claude Sonnet 5.5 or Claude Sonnet 5?
Claude Sonnet 5.5: Input $0.60 / Output $3.00 per 1M tokens; Claude Sonnet 5: Input $0.90 / Output $4.50 per 1M tokens. Rates depend on the billing unit, specification, and usage. Compare matching conditions in the model's detailed pricing; a single rate does not determine the total cost.
What are the key differences between Claude Sonnet 5.5 and Claude Sonnet 5?
Both models share similar capabilities.
Sources
- Introducing Claude Sonnet 5.5
- Claude Sonnet 5.5 model overview
- Introducing Claude Sonnet 5
- Claude Sonnet 5 model overview
Reviewed Oct 2, 2026