Anthropic's Priciest AI Model Struggles to Gain Traction Ahead of IPO

According to the Financial Times, Anthropic's US customers are switching to cheaper alternatives instead of using its most powerful AI model. Data from payment firm Ramp shows that two months after launch, spending on Anthropic's priciest model, Fable 5, is only about 11% of total spending and leveling off, breaking the pattern where businesses default to the strongest model. Analysts say high prices and the fact that older models handle most business needs are key reasons.

This trend could change the business model of frontier AI labs, which have spent billions on training bigger models. An Accel partner says the era when clients only pick frontier models isn't lasting. Demand for Fable 5 is weak, and its adoption is slower than previous models, adding uncertainty before Anthropic's IPO, which could value it at $2 trillion.

Anthropic's annualized revenue in July was $65 billion, below investor hopes, but has grown nearly sevenfold since the start of the year. Meanwhile, OpenAI's GPT 5.6 is priced lower and boosted performance after its July launch, with annualized revenue now over $40 billion. Cheap open-source models from China and elsewhere give customers more options.