Anthropic releases flagship model Claude Opus 4.8

Anthropic announced the release of its new flagship artificial intelligence model, Claude Opus 4.8, on Thursday local time, upgrading from the 4.7 version released last month.

In key benchmark tests such as the agent coding, the new model outperformed OpenAI's GPT 5.5 and Google's Gemini 3.1 Pro in all categories except one. Tests show that the probability of Opus 4.8 missing vulnerabilities when writing code has been reduced by four times compared to Opus 4.7, and it has introduced dynamic workflows for handling large-scale problems.

The standard rate for the new model remains the same as its predecessor, with its fast mode increasing speed by 2.5 times while reducing costs by three times. Anthropic also plans to launch a more powerful Mythos model in the coming weeks. This model has advanced cybersecurity capabilities, which have previously raised concerns among some corporate executives and global leaders.

While the product is rapidly iterating, Anthropic recently completed a huge financing round of $65 billion, with a post-investment valuation reaching $965 billion. This figure has surpassed the valuation of competitor OpenAI, which stands at $852 billion. Just three months ago, the company's valuation was only $35 billion.

External funds are also continuing to pour in, with Google announcing a total investment of up to $40 billion, and Amazon planning to add up to $20 billion in investment. Anthropic previously announced the company's plan to implement an initial public offering by the end of 2026. Market analysis points out that the current AI technology competition has evolved from the early monopoly of OpenAI to a normalized tug-of-war pattern where major giants take turns leading and multiple strong players coexist.