Agent token usage 5× human, but caching discounts cut the real bill to ~2×

智能体 token 用量过线人类、OpenAI 自研芯片跑分出炉、GitHub 发布文档压缩原型

OpenRouter data shows agents consume 7.3T tokens weekly, nominally 5.2× human usage. But 70–85% are cached reads; with ~90% discount, the real bill is roughly 2×. GitHub's Knowledge Compressor prototype halves doc length and claims breakeven at 2,000 reuses, but factoring in caching pushes the median to 5,000+. OpenAI's Jalapeño chip beats Nvidia GB200/GB300 on fixed-length benchmarks, yet lacks AgentX scores for real agent workloads. All three stories share one distortion: prompt caching inflates headline numbers.

Why it matters: Three stories bundled, but the core value is the first: someone finally separated nominal agent token consumption from the caching-discounted real cost, landing at ~2x. The OpenAI chip benchmark and GitHub compression prototype are bonuses but less dense. Cross-source cluster ...