Anthropic launches new artificial intelligence model, promoting superiority in coding

(Patria) - American startup Anthropic has announced the launch of its new generative artificial intelligence model - Claude Sonnet 4.5, which they claim is the best in the world for computer programming, AFP reported.

Anthropic was founded in early 2021 by former OpenAI employees who believed their employer, led by CEO Sam Altman, was not doing enough to control and prevent the potentially harmful effects of its models.

With the support of Amazon, it quickly joined the major players in generative artificial intelligence (AI) who have engaged in a frantic race since the arrival of OpenAI's ChatGPT in November 2022, with new models being released at a rapid pace with increasing capabilities.

Although it lagged behind OpenAI in terms of user numbers and recognition, Anthropic has been considered the best performer in generative AI for computer coding for several months.

This is considered a highly strategic achievement, with programming often cited as the specialty most ripe for disruption and revenue generation by artificial intelligence in the near future.

But OpenAI's latest assistant, GPT-5, launched in early August, has taken the lead in certain AI-generated programming rankings, putting pressure on Anthropic to offer more capabilities in its next offering.

In a key benchmark, Claude Sonnet 4.5, a next-generation language model, can operate autonomously for thirty hours continuously after being assigned a task.

This is a significant leap from Anthropic's most powerful version to date, Claude 4 Opus, which could only operate for seven hours.

These generative AI programs function independently for several hours as they regularly evaluate their own output and autonomously make changes and corrections.

Claude Sonnet 4.5 achieved the highest score when tested by the independent evaluation system SWE-Bench Verified, developed by researchers from Princeton and Stanford universities.

It is also, according to Anthropic, the most advanced model for developing AI agents capable of making real-world decisions for which they have not been trained or specifically programmed.

Anthropic's new release is also the most sophisticated for applications that allow an AI assistant to use a computer as a human would.

At a request written in everyday language, the interface can perform a Google search or update a calendar. Anthropic first offered this functionality in October 2024.

OpenAI launched an equivalent product, Operator, in January 2025.