# OpenAI Jalapeño 芯片 2026 年底内部部署

- 来源：Rohan Paul (@rohanpaul_ai)
- 发布时间：2026-08-30 21:48
- AIHOT 分数：62
- AIHOT 链接：https://aihot.virxact.com/items/cmtfwa3y703h5rooe1ocpc4jx
- 原文链接：https://x.com/rohanpaul_ai/status/2094059690963632455

## AI 摘要

Semianalysis 报道，OpenAI 自研 Jalapeño 芯片将于 2026 年底部署于自有算力。该芯片在约 100 tok/s/user 下实现约 11M tok/s/MW，约为最佳 Blackwell 级配置的 2 倍，且采用单 token 预测（STP）而非多 token 预测（MTP），性能全面超越其他芯片。

## 正文

Semianalysis on OpenAI's new Jalapeño chips that will be deployed inside its own compute by the end of 2026.

"Jalapeño smokes every other chip. All this is done without Multi Token Prediction (MTP), while the other chips on the chart are the best performing configs of each respective SKU, all with MTP"

Jalapeño shifts the entire latency–efficiency Pareto frontier upward: at ~100 tok/s/user it delivers ~11M tok/s/MW, roughly 2× the best Blackwell-class configs at comparable interactivity.

The kicker: that’s STP (Single-Token Prediction) with no speculative decoding, while the competing curves are already using MTP.