Anthropic has released Claude Sonnet 5.5, the second model in its Claude 5.5 family, just days after the launch of the more powerful Claude Opus 5.5. The new model replaces Sonnet 5 as Anthropic’s mid-tier “workhorse”, and its main selling point is not raw power but efficiency: more speed and lower costs for everyday work.
Pricing through the API is unchanged from Sonnet 5, at USD 2 per million input tokens and USD 10 per million output tokens. According to Italian tech site HDblog, those are the same standard rates as OpenAI’s GPT-6 Sol, a sign of how closely the two companies now compete on price.
Where the savings come from
For anyone using AI every day, the price per token is only part of the story. Anthropic says Sonnet 5.5 generates output more than 30 percent faster than Sonnet 5 and typically needs far fewer tokens and fewer tool calls to complete the same work. The result, according to the company, is a cost reduction of up to 30 percent per task for most workloads, even though the rate card has not changed. Anthropic also calls it its fastest Sonnet model to date.
As always, these are the manufacturer’s own figures, and real-world savings will depend on the tasks. Independent testing already shows the nuance: in a test by benchmarking firm Artificial Analysis at maximum effort, Sonnet 5.5 actually came out more expensive per task than Sonnet 5.
Built for well-defined tasks
Anthropic positions Sonnet 5.5 for everyday work with a clear scope: writing and reviewing code, fixing bugs, running well-defined tasks through AI agents, and producing documents, slides and spreadsheets. Early testers, the company says, found it a better collaborator than its predecessor, with clearer writing and speed suited to quick iteration.
The benchmark gains are striking. On Terminal-Bench 4.0, an agentic coding test, Sonnet 5.5 scores 70.6 percent, compared with 10.3 percent for Sonnet 5, and even ahead of the 66.4 percent achieved by the more expensive Opus 5.5. A single score, however, cannot predict how a model will perform in every project.
Tighter safeguards
Because it is more capable, Sonnet 5.5 also comes with stricter protections. It is the first Sonnet model to launch with cybersecurity safeguards similar to those Anthropic uses for its most capable systems. For some higher-risk requests, Anthropic’s apps automatically fall back to Sonnet 5.
Availability
Claude Sonnet 5.5 is available now in the Claude apps and in Claude Code, and for developers via the Claude Platform under the model ID claude-sonnet-5-5. It is also offered through Amazon Web Services, Google Cloud and Microsoft Azure. A third model in the family, Claude Haiku 5.5, aimed at high-volume and cost-sensitive use, is due in the coming weeks.
The launch reflects a wider shift in the AI race. The focus is no longer only on building ever more powerful models, but on bringing part of the capabilities of the most expensive ones into daily work, where speed and running costs matter as much as the quality of the result.
