# Emad Mostaque 预测前沿模型明年将以百倍速度与成本优势一键完成任务

- 来源：Emad (@EMostaque)
- 发布时间：2026-09-04 07:16
- AIHOT 分数：21
- AIHOT 链接：https://aihot.virxact.com/items/cmtm6384k015uro6295pbtqz4
- 原文链接：https://x.com/EMostaque/status/2095652283253993836

## AI 摘要

Emad Mostaque 引用 GPT-6 Astra 在 Artificial Analysis 智能指数上因极高 token 效率处于帕累托前沿、每任务成本低于比其便宜 13 倍的 Gemini 3.8 Flash 的对比，预测明年前沿模型将以比现在快且便宜 100 倍的方式 one shot 完成一切。他认为届时无需监控推理痕迹，是本能式智能，因此预训练中嵌入的内容更加重要。

## 正文

Next year frontier models are going to one shot everything 100x faster & cheaper than they are

No reasoning traces to monitor

Instinctive intelligence

Which is why what you embed in pretraining is even more important

### 引用推文

> cheaty：GPT-6 Astra is on the pareto-frontier of cost efficiency due to being EXTREMELY token efficient. It is in a whole league of it's own. It is cheaper than Gemini ...