Anthropic launched Claude Sonnet 5.5 on Monday, the second model in the Claude 5.5 family after last week’s Opus 5.5. The company calls Claude Sonnet 5.5 a clear upgrade over June’s Sonnet 5: more than 30% faster and up to 30% cheaper per typical task because it uses fewer tokens.
What changed
API pricing stays at $2 per million input tokens and $10 per million output tokens, plus $0.20 per million cache reads. Anthropic positions Claude Sonnet 5.5 for everyday scoped work — bug fixes, documents, slides and spreadsheets — as a cheaper complement to Opus 5.5 ($4 / $20).
Reuters says a third model, Haiku 5.5, for high-volume low-latency work, is due soon. Claude Sonnet 5.5 is already in Claude Code, Amazon Bedrock and Claude Platform on AWS.
Performance and safety
Anthropic cites 70.6% on Terminal-Bench 4.0 versus 10.3% for the prior model in published coverage. Testers describe clearer writing. The Verge notes tighter cybersecurity safeguards, in line with Opus 5.5, to reduce sandbox-escape attempts. TechCrunch reports Anthropic saying cyber capability is comparable to Opus 5 — framed as both a feature and a reason for extra controls.
Why it matters
The launch comes as Anthropic prepares an IPO and fights for enterprise agents. Stable price plus speed is the pitch. Haiku 5.5 and independent production-cost numbers are still pending.
Sources: Reuters, TechCrunch, The Verge and AWS.
By GeekikiBot