---
format: "aidr-story-markdown/v1"
id: "954b449aebf280a0262b626b90e36b0b8d2912bb20c2dac31a9c19679d5e3783"
canonical_url: "https://aidr.today/954b449a?lang=en"
title: "Researchers from Stanford and UC Berkeley Introduce TokenGate: Token-Gated Attention Makes Long-Context Prefill Up to 11.3x Faster"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-10-25T12:12:18.000Z"
category: "Research"
topics: ["inference"]
source_urls: []
summary: "Researchers from Stanford and UC Berkeley introduced TokenGate, a token-gated attention module that accelerates long-context prefill by up to 11.3x and reduces peak memory by up to 3.4x without accuracy loss."
---

# Researchers from Stanford and UC Berkeley Introduce TokenGate: Token\-Gated Attention Makes Long\-Context Prefill Up to 11\.3x Faster

> [Open the canonical story](<https://aidr.today/954b449a?lang=en>)

**Published:** 2026-10-25T12:12:18.000Z
**Category:** Research
**Topics:** inference

## Summary

Researchers from Stanford and UC Berkeley introduced TokenGate, a token\-gated attention module that accelerates long\-context prefill by up to 11\.3x and reduces peak memory by up to 3\.4x without accuracy loss\.

## Sources

_No safe source URL was recorded._