Anthropic positions Claude Sonnet 5.5 for frequent, well-scoped work: fixing bugs, completing everyday coding tasks, and producing polished documents, slides, and spreadsheets. The company says it generates output more than 30% faster than Sonnet 5 and can cost up to 30% less per task. The claimed task savings mainly come from using fewer tokens, not from a lower token price. Read Anthropic’s release notes.

Official image for Claude Sonnet 5.5 发布：单价不变，官方称任务成本降低
Official image from Anthropic; click to open the original source.

Separate list price from task cost

Sonnet 5.5 keeps Sonnet 5’s standard rates: $2 per million input tokens, $10 per million output tokens, and $0.20 per million cached-read tokens. Anthropic says the new model often needs fewer tokens to finish the same task, so the bill for that task may fall. Your workflow is not guaranteed to save 30%. Tool calls, output length, retries, and human corrections all affect the total.

Anthropic reports results on Terminal-Bench 4.0 and GDPval-AA and describes improvements in long-horizon work, image understanding, and collaborative writing. Faster iteration may help with drafts and smaller fixes; Anthropic positions Opus 5.5 for work requiring more complex judgment. Benchmark scores are tied to specific tests and should not replace your own acceptance criteria.

How to try it fairly

  1. Confirm that Sonnet 5.5 is available in your Claude product or API platform; use the official developer documentation for the exact model ID.
  2. Select 10–20 representative tasks and hold the input, tools, permissions, and output format constant for Sonnet 5 and 5.5.
  3. Record first-pass success, edits, response time, and token cost. For code, include test results; for documents, check factual accuracy and formatting.
  4. Validate with real internal examples before rollout instead of treating a demo or leaderboard position as a production guarantee.

Availability and safeguards

Anthropic says Sonnet 5.5 is available through Claude Platform and cloud providers including AWS, Google Cloud, and Microsoft Azure. Because Anthropic considers its cybersecurity capabilities comparable to Opus 5’s, the Sonnet release includes cyber safeguards and fallbacks similar to those used for more capable models. The company says routine software development is unaffected. Check regional availability, model IDs, limits, and prices separately for each platform.