Anthropic launches Claude Opus 5.5 with enhanced speed, lower costs, and expanded usage limits

On Tuesday, the 22nd, Anthropic unveiled Claude Opus 5.5, the latest and most advanced version of its Opus family. According to the company, the new model delivers performance comparable to Claude Fable 5.1 across most tasks, while consuming 40% fewer resources than its predecessor, Opus 5.

Expanded usage limits and improved speed

One of the first changes introduced in the new model is the expansion of usage limits. The company has increased the hourly allowance for Pro, Max, and Team plans, extending it beyond the previous five-hour limit and enabling subscribers to send more messages within this timeframe.

Additionally, users now have access to an extra limit renewal, which can be stored and utilized during periods of high demand, such as busy workdays.

Opus 5.5 also boasts increased speed, with Anthropic claiming it generates responses 30% faster than Opus 5 and completes lengthy tasks with fewer steps. Customers who tested the solution before its launch reported that a programming project previously requiring 38 commands over four days was completed with just 11 commands in three hours.

Clearer text and enhanced reliability

For non-programmers, the most noticeable improvement is in the assistant’s writing style. Anthropic explains that the model’s communication was a major criticism of Opus 5, which often produced lengthy responses filled with technical jargon.

Opus 5.5 prioritizes the most relevant information at the beginning, reduces technical terminology, and more accurately follows user-defined guidelines, such as text length and style.

In an internal reliability test, Anthropic asked Opus 5.5, Fable 5.1, and Opus 5 to draft a report on a company’s quarterly results. An automated checker verified every number and quote, and any fabricated data resulted in the text being rejected. Opus 5.5 passed 16 out of 18 attempts, while the other two models failed all tests.

Advanced capabilities in biology and cybersecurity

Given Opus 5.5’s advanced capabilities in biology and cybersecurity, Anthropic has implemented additional controls for these areas.

In practice, most cybersecurity tasks are now automatically handled by a previous model, Opus 4.8, except for routine programming activities like identifying and fixing flaws in the code itself. Biology researchers can request unrestricted access through a verification program, and cybersecurity professionals will have a similar tool available in the coming weeks.

Price reduction for API-based developers

For developers and companies using Claude via the API, the key update is a cost reduction. The price per million tokens, the billing unit for processing, has dropped by 20%, to $4 for input and $20 for output. Reading cached data has also become 60% cheaper.

Availability and upcoming releases

The model is now available across all Anthropic platforms and on Amazon, Google, and Microsoft clouds. The lighter and more cost-effective Sonnet 5.5 and Haiku 5.5 versions are expected to launch in the coming weeks.