# Epoch AI：GPT-6 Astra 长上下文定价与延迟延续 GPT-5.6 模式

- 来源：Epoch AI (@EpochAIResearch)
- 发布时间：2026-09-11 04:38
- AIHOT 分数：42
- AIHOT 链接：https://aihot.news/items/cmtw0mykk02pbro4v51t5xso1
- 原文链接：https://x.com/EpochAIResearch/status/2098149239146914021

## AI 摘要

Epoch AI 测量显示，OpenAI 新发布的 GPT-6 Astra 在超过 272k 输入 token 后 API 定价上涨，延迟曲线与 GPT-5.6 相似。此前其研究已发现 GPT-5.6 与 Claude 5 的长上下文 TTFT 扩展方式不同：GPT 呈明显二次方增长，Claude 接近线性，对应两者不同的长上下文定价结构。

## 正文

Earlier this week, we showed that GPT-5.6 and Claude 5 exhibit very different long-context TTFT scaling, which corresponds to their different pricing structures at long context lengths.

OpenAI recently released GPT-6 Astra, which shows the same pattern of increased API pricing beyond 272k input tokens. We collected additional latency measurements, which show a similar curvature to that of GPT-5.6 models.

### 引用推文

> Epoch AI：OpenAI GPT-5.6 models and Anthropic Claude 5 models have different pricing structures at long context lengths. GPT model costs increase in price past 272k input...