arena· @arena · X·· 6 小时前AI 评分
Agent Arena 分析显示头部智能体每生成 1 个 token 约处理 93 个输入 token
Arena 对 Agent Arena 逾 2.6 万次会话的分析显示，头部 10 个智能体每生成 1 个 token 约处理 93 个输入 token。作者认为，这解释了各实验室竞相降低缓存输入成本的原因，并以 Opus 5.5 和 Gemini 4 Argon 的真实会话演示缓存机制。
The top 10 agents in Agent Arena process about 93 input tokens for every token they generate, according to our analysis of 26K+ sessions.
This is why labs are racing to lower the cost of cached input.
The video below shows how caching works in practice, with real sessions from Opus 5.5 by @AnthropicAI and Gemini 4 Argon by @GoogleDeepMind.
Dive into the details of the hidden AI price war: article by @DawidGalarowicz below.
来源：arena · x.com