VibeusClaude Sonnet 5.5 is now available, the second model in Anthropic’s... - Vibeus
Claude Sonnet 5.5 is now available, the second model in Anthropic’s 5.5 family. Anthropic positions it as a faster, lower-cost complement to Opus 5.5: strongest on well-scoped everyday work such as fixing bugs and creating polished documents, slides and spreadsheets, with a particular focus on design. Haiku 5.5, aimed at high-volume and cost-sensitive use, is expected in the coming weeks.
The per-token price is unchanged from Sonnet 5: $2 per million input tokens, $10 per million output tokens, $2.50 per million cache writes and $0.20 per million cache reads. Anthropic says Sonnet 5.5 generates output 30%+ faster and usually needs fewer tokens, making it up to 30% cheaper per task in its testing. “Up to” matters: the saving varies by task.
On Terminal-Bench 4.0, which measures complex multi-step command-line work, Sonnet 5.5 scored 70.6% at Max effort, versus 10.3% for Sonnet 5; Opus 5.5 scored 66.4% at Xhigh. In Anthropic’s cost chart, Sonnet 5.5 at Medium—the default in Claude apps—beats Sonnet 5’s best score for less than a tenth of the cost per task. Anthropic cautions that benchmarks capture only part of a model’s ability: Opus 5.5 remains stronger on complex, open-ended work that needs sustained judgment.
For knowledge work, GDPval-AA covers tasks across 44 occupations and nine industries. Sonnet 5.5 scored 1,844, nearly level with Opus 5.5 at 1,846 and about 400 points above Sonnet 5 at 1,449. These scores came from a pre-release deployment with a structured-output bug that has since been fixed; Anthropic expects any effect to be small and says the results may understate Sonnet 5.5’s performance. Other examples include a 55.5% best score on CursorBench 4.0, about two points behind Opus 5.5, and a 90% completion rate in Unity’s multi-step editor and coding benchmark. Unity says it reopens projects and checks results at runtime, so a task counts only when the change works.
Early partner results are also task-specific: Slack reports better offline Slackbot evaluations in fewer steps and with about 14% fewer output tokens; Zendesk reports support tickets processed 20% faster than with the Claude models it currently uses. In Anthropic’s own test, two experts judged the first draft of a 10-slide operating review ready to send as is. These are company-reported evaluations, not guarantees for every workflow.
On safety, an automated audit of roughly 1,850 scenarios found Sonnet 5.5 improved on or matched Sonnet 5 on most measures of alignment, resistance to misuse and honesty. Anthropic notes that evaluations cannot catch every failure. Higher-risk cybersecurity requests visibly fall back to Sonnet 5, while routine software development is unaffected; the biology safeguards are unchanged, most research, education and clinical work is unaffected, though some microbiology or virology requests may be flagged by mistake. Sonnet 5.5 is also the first Sonnet model with classifiers against reasoning-extraction attacks. Its preserved-thinking feature keeps reasoning linked to the account that created it, which can matter when moving conversations between accounts or switching accounts in Claude Code.
Sonnet 5.5 is available across Anthropic’s platforms, including AWS, Google Cloud and Microsoft Azure, with zero data retention. The API model name is claude-sonnet-5-5. If you use it with thinking turned off, Anthropic says to switch to the new between_tools setting first.
