The IPO is imminent, so why did Anthropic's valuation fall after releasing its most powerful new model?

Original | Odaily@OdailyChina)

Author | Wenser (@wenser 2010 )

With its IPO imminent, Anthropic has nonetheless staged a dramatic "model launch, pricing retraction" spectacle.

In the early hours of November 10, Anthropic's Claude announced the release of its latest models, Claude Fable 5.2 and Claude Mythos 5.2, positioned to tackle the hardest technical tasks and said to feature major improvements in coding, reasoning, and vision. Ethan Mollick, a professor at UPenn's Wharton School, declared that the new model "has nearly reached AGI level." But stock pricing documents that subsequently circulated in the market cast a shadow over the celebration, triggering sharp volatility in on-chain prices.

In the battle for the title of "America's first AI large-model stock," Anthropic has already pulled ahead of OpenAI, but whether it can fuel a new AI bull run—or instead burst this round of AI bubble—remains a big question mark.

Claude's New Models Debut: 20% Lower Prices, 30% Faster Output, Fermat's Last Theorem Solved in 30 Hours

According to Anthropic's official blog, the two latest models share the same underlying model. Fable is available to all paid users and has already launched on the Claude platform as well as Amazon Web Services, Google Cloud, and Microsoft Azure; Mythos requires application through a life sciences or cybersecurity verification program. The company says output speed is up to about 30% faster than 5.1.

Claude's follow-up post and the API pricing page are consistent: standard API pricing is $8 per million input tokens and $40 per million output tokens, below the previous version's $10 and $50. The optional fast mode is $16 per million input tokens and $80 per million output tokens, with the company claiming token generation speeds of up to about 2.5x. Additionally, pre-release testing noted that the new model designed three production-ready electronic devices from scratch, solved two unresolved problems related to physics black holes, and completed a formal proof of Fermat's Last Theorem within 30 hours.

According to information from various AI evaluation sites, Claude's latest model is currently the undisputed "No. 1 AI large model."

But product updates and performance gains do not guarantee smooth sailing in the capital markets—for Anthropic, with its IPO imminent, the questions it needs to answer go far beyond this one.

The ANT Pre-Market Pricing Turmoil: With IPO Imminent, Anthropic Withdraws a Pricing Document

The company's official announcement on October 26 confirmed that it had submitted an S-1 to the U.S. Securities and Exchange Commission, planning to list Class A common stock on the Nasdaq Global Select Market under the ticker ANT, with Goldman Sachs and Morgan Stanley as underwriters; at that time, the number of shares to be offered and the price range had not yet been determined. The CNBC live coverage that day listed a total of 22 underwriters. At least 15 calendar days, as required by the SEC, have now passed since the public filing, and this week enters the window in which the company can meet directly with investors.

According to a Wall Street Journal report citing people familiar with the matter, the company still plans to proceed with its IPO this month, targeting a Nasdaq listing and the start of trading on November 19 or 20. Notably, although the market had previously estimated its valuation at close to $2 trillion, amid the recent sell-off in AI stocks, the market's recent expected valuation is about $1.5 trillion, and a further downward revision to $1.2–1.3 trillion cannot be ruled out.

And the vague valuation also led to a "pricing document withdrawal saga" today.

Reuters, citing an email sent to investment banks, reported that Anthropic had at one point floated a plan involving about 867 million shares at $70 to $80 per share, raising up to about $69.3 billion, after which word emerged that the pricing details were inaccurate and would be updated separately. An Anthropic spokesperson declined to comment.

Prior to the IPO, Anthropic's most recent funding round, the Series H, closed on May 28 this year. According to the official announcement, it raised $65 billion at a post-money valuation of $965 billion, led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital.

The $65 billion includes $15 billion previously committed by cloud providers, of which $5 billion is from Amazon, and should not be equated with cash received directly. Anthropic's earlier prospectus also makes clear that the money-burning competition among AI large models is far from over, which is seen as one of the main reasons Anthropic is rushing to push forward its IPO.

On-Chain Perpetuals Offer a "Fair Price" First

After Anthropic's pre-market pricing turmoil, on-chain ANT pre-market pricing also triggered a round of volatility.

On Hyperliquid, the ANT (IO) mark price briefly spiked to around $1,748 in the early hours, and is currently quoted at $1,651. Under that market's pricing method, Anthropic's implied valuation is about $1.64 trillion, down about 12% from the November 4 close of $1,865.4. The instrument's 24-hour trading volume is about $50.96 million, with open interest of about $160 million notional.

In addition, ANT (XYZ), quoted on an expected per-share basis, briefly surged to $105 in the early hours and is currently quoted at $98.5, down about 1.32% over 24 hours. The instrument's 24-hour trading volume is about $38.53 million, with open interest of about $81 million.

Meanwhile, on Lighter, ANTHROPIC pre-market pricing briefly spiked to $1,726 and is currently quoted at around $1,648, only about $3 apart from Hyperliquid's (io) pricing, with 24-hour trading volume of about $5.77 million and open interest worth about $6.16 million—depth far below the two larger channels above.

Overall, Anthropic's pre-market pricing remains disputed, but the valuation range remains above $1.5 trillion. As for the final pricing, the answer will have to wait for the official pricing documents.