Anthropic Launches Claude Opus 5.5, Its Most Powerful Model Yet, At 40% Lower Cost

Anthropic released Claude Opus 5.5, the first model in a new '5.5' family, positioning it as the company's most capable AI system to date while cutting the cost of running it by 40 percent compared with its predecessor, Opus 5.

Performance close to Fable 5.1, at a fraction of the price

According to Anthropic, Opus 5.5 performs at a level similar to Claude Fable 5.1, currently the company's top-tier model, on most tasks, despite sitting a tier below it in Anthropic's lineup. The company said the new model surpassed Fable 5.1 on benchmarks covering agentic coding, knowledge work, computer use, visual chart recognition and multidisciplinary reasoning.

Anthropic described Opus 5.5 as a 'major step up' from Opus 5, citing gains in both performance and safety, and said output generation is now more than 30 percent faster than the previous model. The company added that Opus 5.5 communicates more naturally, placing key information earlier in its responses so it is easier to scan at a glance.

Sharp price cuts across the board

The pricing changes are significant. Opus 5.5 is priced at $4 per million input tokens and $20 per million output tokens, a 20 percent reduction from Opus 5's $5/$25 rate. Anthropic said the combination of lower per-token pricing and greater efficiency brings typical workload costs down by roughly 40 percent overall.

Cache-read pricing saw the steepest cut, falling 60 percent to $0.20 per million tokens. A faster processing mode is also available at a premium, priced at $8 per million input tokens and $40 per million output tokens.

Safety safeguards tightened, not loosened

Unlike Opus 5, which launched with relatively relaxed restrictions after Anthropic found it less capable of exploiting cybersecurity vulnerabilities, Opus 5.5 is being treated more cautiously. Anthropic said the model's biology and cybersecurity capabilities are comparable to those of Claude Mythos 5.1, its most safeguarded system, and it is therefore launching with the same protective measures applied to Fable 5.1.

Those safeguards restrict the model's ability to be used for tasks such as discovering exploits in compiled software or supporting the development of biological weapons. Organisations that need the restricted capabilities for legitimate research can apply for access through Anthropic's new Life Sciences Verification Program.

The model also ships with 'preserved thinking', an anti-distillation safeguard first introduced with Fable 5.1 that is designed to stop bad actors from extracting a model's underlying reasoning through fake accounts at scale. It applies to API accounts created on or after August 31, 2026.

Anthropic said Opus 5.5 was tested before release by external evaluators, including Frontier Design and METR, and performed strongest among its models to date on the company's internal behavioural audit, its most extensive alignment test.

Wider availability, higher usage limits

Claude Opus 5.5 is available immediately across all of Anthropic's platforms, including the Claude Platform, Amazon Web Services, Google Cloud and Microsoft Azure. Developers can access it via the identifier claude-opus-5-5.

Alongside the launch, Anthropic said it is raising five-hour usage limits for subscribers on its Pro, Max, Team and seat-based Enterprise plans, and has issued a rate-limit reset that can be used any time until October 22.

Anthropic said the next tiers in its lineup, Claude Sonnet 5.5 and Claude Haiku 5.5, will follow "in the coming weeks" with similar performance and efficiency improvements.