About the news
Anthropic released Claude Sonnet 5.5 on September 28, 2026 — the second model in the Claude 5.5 family (after Opus 5.5).
Key details:
- Positioned as a faster, lower-cost complement to Opus 5.5
- Best suited for well-scoped everyday tasks: coding, bug fixing, documents, slides, and spreadsheets
- Pricing unchanged from Sonnet 5:
- $2 per million input tokens
- $10 per million output tokens
- More than 30% faster output generation
- Up to 30% lower cost per task (uses fewer tokens and tool calls)
- 1M-token context window and 128K max output
- Model ID: claude-sonnet-5-5
- Available now on Claude API, Amazon Bedrock, Google Cloud, and Microsoft Azure
- Claude Haiku 5.5 expected soon
What matters?
Sonnet 5.5 delivers a major step up for a mid-tier model:
- 70.6% on Terminal-Bench 4.0 (agentic coding)
- vs 10.3% for Sonnet 5
- vs 66.4% for Opus 5.5 (high effort)
- Nearly matches Opus 5.5 on real-world work evaluations (GDPval-AA: 1844 vs 1846)
- Closes much of the practical gap between mid-tier and flagship models for everyday coding and structured tasks
- Costs half as much per token as Opus 5.5 while being faster and more efficient than Sonnet 5
- First Sonnet model to include advanced cybersecurity safeguards previously reserved for Anthropic’s top models
What to watch next?
- Release of Claude Haiku 5.5 (high-volume, cost-sensitive workloads)
- How quickly teams shift everyday production work from Opus 5.5 to Sonnet 5.5
- Independent third-party benchmarks beyond Anthropic’s numbers
- Any updates to safety classifiers or effort controls as usage scales
- Anthropic’s ongoing product cadence amid its broader company plans
