Tokens Too Cheap to Meter
What happened
作者 jyn 用多张图表说明，AI 推理成本每年都在以数量级的速度下降。GPU 能效大约每两年翻一番，到 2026 年，完成同样任务的模型成本比 2025 年底便宜了两个数量级。像 vLLM 这样的推理引擎，每年还能额外带来 10% 到 50% 的吞吐量提升。文章判断，一两年内大模型会像水电一样成为计算基础设施，三到六年内，消费级硬件就能在本地跑通当前...
Coverage
Follow the reports to see the story from different sides.
- Hacker News front pagePickTokens Too Cheap to Meter
jyn argues with multiple charts that AI inference cost is dropping by orders of magnitude each year. GPU efficiency doubles roughly every two years, and per-task model cost in 2026 is two orders of magnitude cheaper than end of 2025. Inference engines like vLLM add 10%–50% throughput gains annually. The author expects LLMs to become computing infrastructure within 1–2 years, and frontier-quality local models on commodity hardware in 3–6 years. The post doesn't cite specific dollar figures, but the trend lines are stark.