According to CoinMeta, the strongest and most expensive Fable model has been on the market for over two months, but corporate customers have not flocked to it. Corporate spending data shows that this model accounts for only 6% of the total usage of the Anthropic model, with a cost proportion of just 11.4%. The input and output prices for this model are $10 and $50 per million token, respectively, which is twice that of the Opus model. The company itself positions this model as "close to the forefront of Fable intelligence, but at half the price." Enterprises have begun to vote with their wallets, and the Opus model, launched at the end of July, has already surpassed the Fable model in terms of corporate spending. In contrast, the GPT-5.6 model now accounts for 25% of corporate spending in OpenAI. Previously, enterprises would usually pursue the latest and strongest models, but this time there has been a clear reversal in trend. Accel, one of the partners, pointed out that most people do not necessarily need to use the most cutting-edge models all the time; if model capabilities continue to improve while prices double, enterprises may not be willing to pay for them.