Anthropic releases Claude Opus 4.5, with comprehensive improvements in programming performance

On Monday, Anthropic launched the latest version of its flagship AI model, Claude Opus 4.5, claiming that this model is stronger in software engineering than previous versions and can better execute automated programming and office tasks. Analysts say this is another move by Anthropic to compete for enterprise clients in the race against OpenAI and Google.

Claude Opus 4.5 is the third major model released by Anthropic in two months, once again showcasing the rapid pace of development in the AI industry. The company launched Claude Sonnet 4.5 at the end of September and then released Claude Haiku 4.5 in October.

Anthropic stated that Claude Opus 4.5 can autonomously fix programming errors without user intervention and is designed to better execute complex multi-step tasks on users' computers and the internet.

Alex Albert, head of developer relations at Anthropic, told the media:

“In the tasks we truly care about, this is the smartest model in the world.”

“Our theme is to push forward at an extremely high speed and continuously release the best models we can.”

Claude Opus 4.5 will be launched in all regions and will become the default model for the entire line of Anthropic Pro, Max, and Enterprise products.

New Model's Programming Capabilities Stand Out

Anthropic stated in its blog that this new model scored higher than Google Gemini 3 Pro and OpenAI's GPT 5.1 on the widely used programming benchmark SWE-Bench Verified.

The new model is "significantly stronger" in handling everyday tasks. In terms of "agentic coding," Claude Opus 4.5 has also reached an industry-leading level, outperforming Gemini 3 Pro and OpenAI's GPT-5.1 according to the results of the software capability assessment set SWE-bench Verified.

According to Scott White, head of Claude AI model products, Anthropic's new model has reached a new programming milestone in a sense. Opus 4.5 is the first model to score higher than all company engineering job applicants in a challenging internal "home engineering task" test. White did not disclose the specifics of this task but mentioned that it is an assessment task that requires qualified candidates to spend several hours completing, and the task itself also utilizes Anthropic's Claude model White stated to the media:

"Now, it has reached a turning point, and we must rethink how to assess software engineering capabilities."

White mentioned that the ideal users of Claude Opus 4.5 include professional software developers, financial analysts, consulting consultants, and accountants, among other knowledge workers. He added that those who are "eager to enhance their creativity, create new products, and expand their professional capabilities" will also find this model very useful.

He stated that the new model can better handle tasks such as financial analysis, creating presentations, and spreadsheets. In addition, Opus 4.5 is more suitable for back-and-forth collaboration with users, rather than just generating a rough draft for users to refine on their own.

Anthropic will also provide Opus 4.5 to enterprise customers and its premium Max subscription users within Microsoft Excel. The chat feature in Excel allows users to instruct the Claude chatbot to perform tasks such as editing spreadsheets. Previously, this feature was only available to invited testers.

Other Product Updates

In addition to the model release, Anthropic also announced a series of other product and feature updates on Monday.

The company stated that its browser extension Claude for Chrome (which allows Claude to perform operations across different browser tabs) will be available to all Max users. Claude for Excel (which can understand and edit spreadsheets) will also be fully available to all Max, Team, and Enterprise users.

Anthropic will also introduce Claude Code to desktop applications and add new features to its developer platform.

Leading Adoption Rate in Enterprise Programming Field

Anthropic was founded in 2021 in San Francisco by former OpenAI employees and currently has over 300,000 enterprise customers using its models to streamline workflows. Particularly in the field of computer programming, the company has become one of the market leaders. Microsoft and NVIDIA announced last week a multi-billion dollar investment in Anthropic, raising its valuation to approximately $350 billion.

The company's most well-known products are a series of AI models named Claude. They differentiate different generations by continuously increasing the numbering, but the largest model in the series is usually referred to as Opus, designed for advanced reasoning and complex problem-solving; the medium-sized model is called Sonnet, and the smallest is called Haiku, both primarily optimized for speed and efficiency. The last Opus model released by Anthropic was in August, named Claude Opus 4.1.

However, Anthropic faces fierce competition from OpenAI and Google. Google released Gemini 3 last week, which performs better in tasks such as coding.

In July of this year, a report from Menlo Ventures pointed out that Anthropic currently leads in enterprise-level AI adoption rates, holding a 32% market share. OpenAI ranks second with 25%, nearly halved compared to two years ago; Google holds 20%, and Meta ranks fourth with 9%