![[535c2078-8669-4757-a34f-744947946c6e.mp3]]

NVIDIA发布Llama Nemotron Nano VL 8B V1，多模态轻量AI模型上线，支持图像视频文本高效解析

简报：

2025年6月，NVIDIA正式发布Llama-3.1-Nemotron-Nano-VL-8B-V1视觉语言模型，该模型基于Llama-3.1架构，拥有8B参数，支持图像、视频与文本多模态输入，具备高质量文本输出及图像推理能力。
该模型在OCRBench V2测试中成绩居首，特别适用于文档智能、图像摘要、光学字符识别等场景，可部署于云端及边缘设备，并通过AWQ量化技术降低硬件门槛。
Llama-3.1-Nemotron-Nano-VL-8B-V1支持128K上下文长度，开放开源，已在Hugging Face平台上线，为开发者和企业级自动化文档处理、知识管理等应用提供低成本、高效率的多模态AI解决方案。

相关链接：

NVIDIA Releases Llama Nemotron Nano VL 8B V1: Multimodal Lightweight AI Model Launched, Supporting Efficient Image, Video, and Text Analysis 🧠

Briefing:

In June 2025, NVIDIA officially launched the Llama-3.1-Nemotron-Nano-VL-8B-V1 vision-language model. Based on the Llama-3.1 architecture, this 8-billion-parameter model supports multimodal input including images, videos, and text, offering high-quality text output and image reasoning capabilities.
The model achieved top scores in OCRBench V2 tests, making it particularly suitable for scenarios like document intelligence, image summarization, and optical character recognition. It can be deployed on both cloud and edge devices, with hardware requirements lowered through AWQ quantization technology. 🚀
Llama-3.1-Nemotron-Nano-VL-8B-V1 supports a 128K context length and is open-source, now available on the Hugging Face platform. It provides a low-cost, high-efficiency multimodal AI solution for developers and enterprise-level applications such as automated document processing and knowledge management. ✨

简报：

OpenAI宣布为ChatGPT新增MCP（Model Context Protocol）支持，使AI模型能与GitHub、SharePoint、Gmail、Dropbox、Outlook等多种企业内部数据源双向链接，提升企业知识检索与管理效率。
ChatGPT上线会议记录模式，支持自动转录会议内容、提炼要点、生成任务分配和代码片段，目前已面向ChatGPT Team的macOS用户开放，未来将拓展至更多用户群体。
这两项新功能进一步推进ChatGPT在企业级应用与团队智能协作领域的作用，有助于提升办公自动化与数据利用率。

相关链接：

Brief:

OpenAI has announced new MCP (Model Context Protocol) support for ChatGPT, enabling the AI model to establish two-way links with various enterprise internal data sources such as GitHub, SharePoint, Gmail, Dropbox, and Outlook. This aims to enhance corporate knowledge retrieval and management efficiency.
ChatGPT has launched a meeting recording mode, supporting automatic transcription of meeting content, extraction of key points, generation of task assignments, and code snippets. This feature is currently available to macOS users of ChatGPT Team and will be expanded to more user groups in the future. 📝
These two new functions further advance ChatGPT's role in enterprise applications and intelligent team collaboration, contributing to improved office automation and data utilization. 💡

简报：

相关链接：

Brief:

Amazon has announced plans to invest approximately $10 billion 💰 in North Carolina to expand its data center infrastructure, supporting the development of artificial intelligence (AI) and cloud computing technologies.
This move is expected to create at least 500 high-skill jobs and thousands of additional employment opportunities related to the data center supply chain.
North Carolina's geographical location, lower operating costs, and strong state government support for the tech industry make it an ideal choice for Amazon's data center expansion.
This investment will help enhance Amazon's infrastructure competitiveness in global cloud computing and AI, while also boosting the local economy and technology industry development.

简报：

相关链接：

Briefing:

Tesla announced that its AI-powered autonomous taxi service, Robotaxi, will commence its first phase of trial operations in Austin, Texas, USA, on June 12, 2025. 🤖
Initially, the trial fleet will consist of only 10 Model Y vehicles, operating exclusively within the safest areas of Austin, and will be accessible only to selected invitees, not available for general public hailing. 🗺️
During the initial operational phase, remote human intervention will provide support, with engineers in California remotely taking over in complex road situations to enhance operational stability. 🚦
Tesla plans to expand the fleet to thousands of vehicles within a few months and gradually extend the service to other cities.