Anthropic · Claude Sonnet
Claude Sonnet 5.5
The current Sonnet for writing, coding and tool workflows; test your existing integration before migrating from Sonnet 5.
Start here
- What this helps you do
- Decide whether Claude Sonnet 5.5 is the right model for a job you have, or one to leave alone.
- What you leave with
- A plain read of what Claude Sonnet 5.5 is for, what to avoid giving it, and the facts behind that.
- What to check before trusting it
- Adaptive thinking defaults to high effort; between_tools replaces disabled at high effort or below.
Treat this as a migration candidate, not a tested quality win. Compare the same job, token use, failures and human corrections before routing work to it.
Use it for
- Professional writing and analysis
- Coding and tool-using workflows
- Long-context work with human verification
Do not swap model IDs without checking thinking, tool settings and response parsing. Human review still owns correctness and actions.
Effort without the jargon
Use adaptive thinking by default and calibrate effort on the real task.
For no up-front thinking, use between_tools at high effort or below and test response handling.
Advanced recordCost, limits, privacy and operations
Start with a small migration trial and measure the human checking work as well as the token bill.
- API cost
- Officially disclosedAPI US$2 input and US$10 output per million tokens; cache reads US$0.20, 5-minute writes US$2.50 and 1-hour writes US$4. Batch input and output receive a 50% discount.
- Context window
- Officially disclosed1M tokens (provider wording)
- Maximum output
- Officially disclosed128K tokens (synchronous API)
- Knowledge cutoff
- Officially disclosedJune 2026
- Recorded inputs
- Text · Images
- Recorded output
- Text
Data handling is a product decision
Depends on surfaceRetention depends on API features, the agreement and platform. Verify that boundary before sending sensitive work.
Operational constraints
- Adaptive thinking defaults to high effort; between_tools replaces disabled at high effort or below.
- Forced tool use and manual thinking budgets are unsupported; non-default sampling parameters and assistant prefill return errors.
- Pass thinking blocks unchanged and keep conversations append-only; blocks are tied to the model and conversation.
- Check computer-use toolset and supported advisor migration. Text between tool calls arrives in thinking blocks, so test visible progress and response parsing.
Record details
- Evidence readiness
- informational
- Guidance confidence
- moderate
- Evidence revision
- claude-sonnet-55@v1
Unknown means the cited official record does not disclose a safe value. It is not an estimate. Prices are provider list prices in USD where stated and can change before this record's review date.
Keep the evidence separate
Anthropic lists Sonnet 5.5 as Active (latest), released 28 September 2026, and recommends it for migration from Sonnet 5.
No sufficiently useful independent note has been added yet. That absence is not filled with our guess.
No Human Bit result is claimed; use this as evaluation guidance only.
Not scheduled
No Human Bit result is claimed for this model yet.