PulseAugur
实时 10:07:15
English(EN) 12 million tokens, linear cost: Subquadratic's bet against the attention tax

Subquadratic 推出具有线性扩展架构的 1200 万 token 上下文窗口

Subquadratic 是一家拥有 11 名博士研究员的初创公司,已推出一款采用其 Subquadratic 选择性注意力(SSA)架构的新模型,该架构声称可以随上下文长度线性扩展。这项创新实现了 1200 万 token 的上下文窗口,旨在克服 LLM 中传统密集注意力机制的二次成本限制。早期基准测试显示,在 MRCR v2SWE-Bench 等任务上,其性能与 GPT-5.5Claude Opus 等模型相当,且推理速度显著更快。 AI

影响 上下文长度的计算和内存线性扩展可以显著降低 RAG 和代理分解的成本并提高投资回报率。

排序理由 一家初创公司发布了一款采用新颖架构的新模型,并提供了基准测试结果。[lever_c_demoted from research: ic=1 ai=1.0]

在 dev.to — LLM tag 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

Subquadratic 推出具有线性扩展架构的 1200 万 token 上下文窗口

本文如何被排名

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
一家初创公司发布了一款采用新颖架构的新模型,并提供了基准测试结果。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
model release, paper
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
116 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准

报道来源 [1]

  1. dev.to — LLM tag TIER_1 English(EN) · Andrew Kew ·

    1200万token,线性成本:Subquadratic的注意力税反击战

    <p>The quadratic attention problem has quietly shaped everything you've built with LLMs. RAG pipelines, agentic decomposition, hybrid architectures — these aren't the natural shape of AI systems. They're workarounds. Doubling the context quadruples the compute, so everyone stoppe…