English(EN) This AI Model Has 2.8 Trillion Parameters. It Only Uses 104 Billion at a Time.

Moonshot AI发布Kimi K3.1 Pro，拥有2.8T参数，每次推理使用104B

作者 PulseAugur 编辑部 · [1 个来源] · 2026-11-13 04:00

Moonshot AI已发布其Kimi K3.1 Pro模型的权重，该模型拥有2.8万亿参数。这个大型模型利用稀疏注意力机制和专家混合（mixture-of-experts）方法，使其在每次推理任务中仅激活1040亿参数。这种效率对于管理如此庞大AI的计算需求至关重要。 AI

影响此次发布展示了在超大型模型中高效利用参数方面的进展，可能影响未来大规模AI的发展。

排序理由前沿实验室模型发布，附带系统卡。[lever_c_demoted from frontier_release: ic=1 ai=1.0]

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

本文如何被排名

Signal score: 83 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Significant; 前沿实验室模型发布，附带系统卡。[lever_c_demoted from frontier_release: 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, infra; 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: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

Towards AI TIER_1 English(EN) · Pop Alexandru · 2026-11-13 04:00

这款AI模型的参数量为2.8万亿，但每次仅使用1040亿。

<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://pub.towardsai.net/this-ai-model-has-2-8-trillion-parameters-it-only-uses-104-billion-at-a-time-c614fbbf1433?source=rss----98111c9905da---4"><img src="https://cdn-images-1.medium.com/max/738/1*stlTll4CKw3p…