11-19 更新 实时
首页/ 深度/ 正文

腾讯发布 Hy4 flash 模型权重,支持本地部署

深度 来源:TechNode 2026-11-19 13:43 0 阅读 机器翻译
腾讯发布 Hy4 flash 模型权重,支持本地部署

Tencent 于周三发布了其 Hy4 flash AI 模型的权重,而两天前该模型才通过其应用和云 API 上线。开发者可以下载标准模型、FP8 版本以及用于进一步训练的基础模型。此次发布兑现了 Tencent 此前关于计划公开该模型权重的声明。

Hy4 flash 支持文本和图像输入,上下文窗口最高可达 100 万 Token。它拥有约 2240 亿参数,每个 Token 约激活 120 亿参数,面向研究、文档准备和表格处理等日常生产力任务。Tencent 已在 Tencent Hy 许可下,随权重一并发布了部署材料。[Tencent]

译文由英文原文机器翻译,术语按行业惯例统一,可在文末展开原文核对。

查看英文原文(Tencent releases Hy4 flash model weights for local deployment)

Tencent released the weights for its Hy4 flash AI model on Wednesday, two days after launching it through its apps and cloud API. Developers can download the standard model, an FP8 version and a base model for further training. The release makes good on Tencent’s announcement that it planned to make the model’s weights available.

Hy4 flash supports text and image inputs and a context window of up to one million tokens. It has approximately 224 billion parameters, with around 12 billion activated per token, and targets everyday productivity tasks such as research, document preparation and spreadsheet work. Tencent has published deployment materials alongside the weights under its Tencent Hy License. [Tencent]

正文保留英文原文,标题与摘要由机器翻译。

本文由采集器自动同步

原文发布于 TechNode(Tencent releases Hy4 flash model weights for local deployment),标题、摘要与正文由机器翻译,术语按行业惯例统一,未经人工逐句校订。

阅读原文 ↗