1. This might mean that Ilya was wrong when he predicted that we’re “moving from scaling the recipe (pre-training) to scaling the thought process (inference)”, if training at scale resulted in a dramatic improvement
2. The costs for other labs to replicate it, means the present is probably the most access most of the public will have to the most advanced models. From this point the cost will diverge into K curves, where the average consumer uses open source level models.

Three weeks ago there were rumors that one of the labs had completed its largest ever successful training run, and that the model that emerged from it performed far above both internal expectations and what people assumed the scaling laws would predict. At the time these were

Andrew Curran

Twitter

1. 如果大規模訓練能夠帶來顯著的性能提升，那麼伊利亞之前關於“我們正在從規模化配方（預訓練）轉向規模化思維過程（推理）”的預測可能就錯了。

2. 其他實驗室複製該模型的成本意味著，目前可能是大多數公眾能夠接觸到最先進模型的最佳時機。從現在開始，成本將呈K曲線式增長，屆時普通消費者將使用開源級別的模型。

Anthropic Claude Code 負責人 Boris Cherny 宣布，自 2026 年 4 月 […]
〈Anthropic 訂閱 Claude Code 封殺龍蝦 OpenClaw！往後第三方工具僅能付費額度〉這篇文章最早發佈於動區BlockTempo《動區動趨-最具影響力的區塊鏈新聞媒體》。

Anthropic 訂閱 Claude Code 封殺龍蝦 OpenClaw！往後第三方工具僅能付費額度

全球最活躍的比特幣買家正以接近歷史最高水平的速度買入。但這還不夠。

CryptoQuant 的一份週報顯示，截至 3 月底，過去 30 天的比特幣表觀需求為負 63,000 $BTC。

五個數據來源對比特幣市場給出了相同的結論：它正在從內部逐漸萎縮。

本輪加密市場正處於一個重要的過渡階段。比特幣在經歷數月回調後，近期出現小幅反彈，並重新測試此前失守的關鍵支撐區間。但整體來看，市場修復基礎仍不穩固，趨勢層面依然偏脆弱。宏觀壓力持續累積、流動性邊際趨弱，同時關鍵政策事件臨近，使得市場定價邏輯正在發生變化。進入4月，市場雖存在季節性偏強預期，但歷史表現分化明顯，行情更可能由流動性、倉位結構與宏觀催化共同驅動，而非簡單的週期重複。
支撐反覆驗證：趨勢脆...