Anthropic launches Claude Opus 5.5 with 40% lower costs

Anthropic has launched Claude Opus 5.5, a new flagship AI model that it says is faster and 40% cheaper to run on typical workloads than Claude Opus 5. The model was announced on Tuesday 22 September and is the first in Anthropic’s Claude 5.5 family.

According to Anthropic, Opus 5.5 produces output more than 30% faster than Opus 5 and performs at the level of Claude Fable 5.1 on most work. The company attributes the claimed overall saving to lower token prices as well as reduced token use per task.

Lower API prices, but no confirmed UK subscription cut

API input tokens cost $4 per million, down from $5 for Opus 5, while output tokens cost $20 per million rather than $25. Cache reads are priced at $0.20 per million tokens, compared with $0.50 previously, and cache writes fall from $6.25 to $5.

Anthropic also offers a Fast mode in Claude Code and on the Claude Platform, which it says can run at up to 2.5 times the speed. That version is priced at $8 per million input tokens and $40 per million output tokens.

The company has not confirmed an Opus 5.5-specific UK subscription price. Its advertised 40% saving refers to typical token-billed workloads, not a blanket reduction on Claude consumer plans.

Available now for paid Claude users

Opus 5.5 is available through the Claude Platform under the identifier claude-opus-5-5, as well as Amazon Web Services, Google Cloud and Microsoft Azure or Microsoft Foundry. It is also available to Claude Pro, Max, Team and Enterprise users.

Anthropic says it has increased five-hour usage limits for Pro, Max and Team plans, and added a rate-limit reset that subscription users can save for when they need it. Claude Sonnet 5.5 and Claude Haiku 5.5 are due in the coming weeks.

Performance and safeguards

Anthropic’s published tests put Opus 5.5 ahead of Opus 5 across coding, automation and computer-use benchmarks. These are company-published or partner-reported evaluations rather than an independent overall ranking, and Anthropic says most results used adaptive thinking at maximum effort.

The company says external evaluators METR and Frontier Design tested the model before release. As Anthropic considers Opus 5.5 comparable with Claude Mythos 5.1 in biology and cybersecurity, it has applied stronger safeguards, including restricted access to many cybersecurity tasks and verification programmes for some users.

For ordinary buyers, the immediate change is more capability within eligible Claude plans rather than a cheaper monthly bill. The larger effect may come later: if the claimed efficiency holds up, UK coding tools and business services using Claude could become less expensive to operate.