Anthropic launches Opus 5.5 with 40 percent lower processing costs

Anthropic has introduced Claude Opus 5.5, a new artificial intelligence model that the company describes as a significant advancement over its predecessor. The latest release is designed to reduce operational expenses, with Anthropic stating that it costs 40 percent less to run than Opus 5 on typical workloads. In a statement issued on 22 September, the company noted that early testers observed substantial performance improvements on complex tasks, positioning the new model as the leading option in its current lineup.

The model is now accessible through the Claude Platform and major cloud infrastructure providers, including Amazon Web Services, Google Cloud and Microsoft Azure. Anthropic emphasised the safety profile of Opus 5.5, noting that it underwent testing by external evaluators prior to release. The company reported that the model performed stronger than other systems on its automated behavioural audit and includes the safeguards developed for its most capable models. Additionally, Anthropic highlighted improvements in communication capabilities, asserting that the writing produced by Opus 5.5 is more natural, clearer and easier to follow than that of the previous version.

Regarding specific capabilities, Anthropic stated that Opus 5.5 leads in agentic coding, computer use and knowledge work on its internal benchmarks. However, the company cautioned that at these levels of capability, benchmark margins have become a less reliable guide to real-world differences. It noted that in its own usage, the gap between Opus 5.5 and Claude Fable 5.1 is narrower than the scores suggest. This launch occurs against a backdrop of increasing industry interest in open-weight AI models, including those from Chinese developers, driven by the high costs of proprietary US models.

The release of the lower-cost model also coincides with reports that Anthropic is preparing for an initial public offering. It was reported on 18 September that the company plans to launch its IPO in November, a date later than the October debut many investors had anticipated. This timing may allow the company to share its third-quarter financials, which are expected to show strong results. Furthermore, reports indicated that Anthropic was considering the launch of a new AI model before going public to counter the momentum gained by rival startup OpenAI following the debut of its GPT-Astra.