GLM-5.3 前阵子以"Ox Alpha"的匿名身份上线时，就已经是 OpenCode 和 OpenRouter 双平台调用量最大的模型。今天智谱给它加了个更快的版本，名字也直白——GLM-5.3-FlashX，最高 200 tokens/s。 背后的一张牌值得单独说。上一代 GLM 上线时，团队是在 10 万张国产芯片组成的集群上把推理服务从零搭起来的，当时的难点是内...
本文为自动采集的行业资讯。原文来源：www.oschina.net
This article was automatically collected from industry news feeds. Source: www.oschina.net
Build a multimodal RAG with Gemini Embedding 2: search text, images, PDFs, video, and audi
NASA 和 IBM 开源了首个月球模型 NASA-IBM Lunar Foundation Model，模型权重与代码向全球研究者开放，以供下载、微调和试验。这是首批专门面向月球
OpenAI introduces the Australian Youth Safety Blueprint, a six-pillar roadmap for safer AI
Google said Gemini had "acted appropriately" by ending each hack immediately.
An interactive directory that matches GPUs, Macs, and edge devices to local AI models, plu
Fine-tuning a 350M Model for Better Structured Outputs in 100 GRPO Steps本文为自动采集的行业资讯。原文来源：
