I've gone down the rabbit hole on GLM 5.2 and the Chinese open-source models are functionally indistinguishable from the world's best closed frontier models for most users.
我深入研究了 GLM 5.2，發現對大多數用戶而言，中國的開源模型在功能上與世界上最好的封閉前沿模型幾乎沒有差別。 I wonder how long this takes the market to start pricing this in, maybe after the anthropic and openAI IPOs, later this year? 我想知道市場還要多久才會開始將這一點納入定價，或許是在 Anthropic 和 OpenAI 上市之後，今年晚些時候？ Technology is structurally deflationary with inflationary spikes, and it's a race to zero on inference costs. 技術本質上是通縮性的，伴隨著通膨的尖峰，而推理成本正在競相趨零。 I believe at some point we will get an open-source-driven deflationary shock, that's when the fed pivots but with tech, each cycle happens quicker. 我相信總有一天我們會迎來開源驅動的通縮衝擊，那時聯準會將會轉向，但隨著科技發展，每個循環都來得更快。
I've gone down the rabbit hole on GLM 5.2 and the Chinese open-source models are functionally indistinguishable from the world's best closed frontier models for most users.
我深入研究了 GLM 5.2，發現對大多數用戶而言，中國的開源模型在功能上與世界上最好的封閉前沿模型幾乎沒有差別。
I wonder how long this takes the market to start pricing this in, maybe after the anthropic and openAI IPOs, later this year?
我想知道市場還要多久才會開始將這一點納入定價，或許是在 Anthropic 和 OpenAI 上市之後，今年晚些時候？
Technology is structurally deflationary with inflationary spikes, and it's a race to zero on inference costs.
技術本質上是通縮性的，伴隨著通膨的尖峰，而推理成本正在競相趨零。
I believe at some point we will get an open-source-driven deflationary shock, that's when the fed pivots but with tech, each cycle happens quicker.
我相信總有一天我們會迎來開源驅動的通縮衝擊，那時聯準會將會轉向，但隨著科技發展，每個循環都來得更快。