The AI sector has seen sharp volatility of late, yet Anthropic is still pressing ahead with its IPO at full speed. On the eve of going public, the company rolled out two new-generation frontier models, Claude Fable 5.2 and Claude Mythos 5.2, with a focus on sharpening coding, reasoning and vision capabilities while optimizing speed and cost control. The two models share the same underlying foundation but differ in their safety safeguards and scope of access. This marks Anthropic's first launch of new frontier models since its CEO called in September this year for a slower pace of AI development; a relevant executive said the call was about keeping development under control, not about pausing R&D. Seven tech giants spanning core links of the AI industry chain, including Nvidia and Google, plan to invest a total of US$26 billion through private placements at the IPO issue price in support of Anthropic's listing. The issuance terms remain subject to change; previously disclosed pricing information has been voided, and the number of shares, the price range and the size of the fundraising are yet to be confirmed. Weighed down by factors such as the sell-off in AI concept stocks, related policies and doubts over return on investment, the company's listing valuation has dropped sharply from the previous expectation of more than US$2 trillion, with the market-expected valuation range now lowered to between US$1.1 trillion and US$1.5 trillion. The company's shares may begin trading as early as November 19 or 20, completing the listing ahead of Thanksgiving. Financial data shows that Anthropic's revenue is growing rapidly, but computing costs remain high and losses are sizable, with signed agreements related to computing power and infrastructure procurement totaling nearly US$900 billion.