4 家媒体·5 篇报道·历时 10h
Kimi发布Kimi K2.5,开源视觉SOTA代理模型
中国初创公司Moonshot AI推出Kimi K2.5,这是一款基于约15万亿混合视觉与文本标记训练的开源模型,能够同时处理文本、图像和视频,内置编码代理,并可自行指挥最多100个子代理的群体,在多项测试中超越开源竞争对手。
本文综合以下全部报道写成，而非依据任何单一来源。
本页由机器翻译生成，原始报道为其原文语言。
各家媒体如何报道
- Hacker News·Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model
- TechMeme·Chinese startup Moonshot updates its Kimi model, saying K2.5 can process text, images, and videos simultaneously and beats its open-source peers in some tests (Bloomberg)
- TechMeme·Moonshot says Kimi K2.5 builds on K2 with "pretraining over ~15T mixed visual and text tokens" and "can self-direct an agent swarm with up to 100 sub-agents" (Kimi)
- TechCrunch·China’s Moonshot releases a new open-source model Kimi K2.5 and a coding agent
- VentureBeat·Moonshot AI debuts Kimi K2.5, most powerful open source LLM beating Opus 4.5 with swarm of parallel agents