Aravind Srinivas· @AravSrinivas · X·· 2 小时前AI 评分47
AI 导读
决策模型每百万输入 token 仅 1 美分（这是我们为 pplx-decider-flash 提供的定价），意味着 1000 万亿输入 token 最多只需花费 $10,000。我们还将进一步优化推理和（输入的）缓存。https://docs.perplexity.ai/docs/decisions/quickstart
正文
At 1 cent per million input tokens for decision models (the pricing we offer for pplx-decider-flash), you basically spend at best $10,000 for a quadrillion input tokens. We're going to optimize inference and caching (of inputs) even further. https://docs.perplexity.ai/docs/decisions/quickstart
来源：Aravind Srinivas · x.com