Back to News

Tencent Hunyuan unveils Hy4 LLM with vision, computer-use, open-source weights

#tencent#hunyuan#hy4#llm

On November 4, 2026, Tencent Hunyuan released Hy4, its latest large language model, marking the first time the Hy series enters the top tier of open-source models. Hy4 introduces vision understanding and computer-use capabilities for the first time, with significantly expanded data and training scale compared to Hy4 preview, and improved instruction following, long-context comprehension, and long-horizon agent performance. The model weights and a system card are scheduled to be open-sourced soon, with a focus on responsible open-sourcing addressing network and biological safety risks.

Coverage timeline

  1. 腾讯混元 BlogHy LLM
    Introducing Hy4

    @@@coverCard type = "video" date = "November 4, 2026" media_src = " https://hy-model-ap-prod-1258344703.cos.ap-guangzhou.myqcloud.com/llm-blog/default/a951c950/2026110219294526_20bbdf2a313c4cad81b3fdb22932e7ff.mov?q-sign-algorithm=sha1&q-ak=AKIDRl074nOsGdJ9zjMsCRWP3ShmgS3VtX4S&q-sign-time=1793618987;2108978987&q-key-time=1793618987;2108978987&q-header-list=&q-url-param-list=&q-signature=7b331fb3eec1fdd37e468ac2348a8fdfd77b1406 " poster=" https://hy-model-ap-prod-1258344703.cos.ap-guangzhou.myqcloud.com/llm-blog/default/a951c950/2026110219302736_cb6bad8aa277cf9d40dc98cfa2d4ad3a.png?q-sign-algorithm=sha1&q-ak=AKIDRl074nOsGdJ9zjMsCRWP3ShmgS3VtX4S&q-sign-time=1793619030;2108979030&q-key-time=1793619030;2108979030&q-header-list=&q-url-param-list=&q-signature=150aa51ca6edd06bbe479a9f85ee8bab9b9ade49 " image_alt = "hy4PreviewBg" title = "Introducing Hy4" aspectRatio = "" mediaWidth = "100%" mediaMarginTop = "32" mediaMarginBottom = "64" videoLoop = 0 @@@ @@@space height = 16 @@@ @@@buttonGrou

  2. 机器之心新闻资讯
    腾讯混元发布 Hy4，跻身最强开源模型之列

    11 月 4 日，腾讯混元发布最新一代大语言模型 Hy4。随着模型能力的再次全面跃升，Hy 首次跻身最强开源模型之列。Hy4 将于近期开源模型权重，发布模型系统卡片，并积极推进更负责任的开源，重点关注网络和生物安全等风险。 相比于 Hy4 preview，Hy4 在预训练和后训练上显著扩展了数据和训练规模，首次引入视觉理解能力，并大幅提升了指令遵循、长文理解、长程智能体等综合效果。 通过视觉理解、推理和工具调用能力，Hy4可以完成细粒度图像处理、复杂图表分析和数学视觉推理等任务。 Hy4 也首次支持电脑操作（Computer Use），可以结合命令行和图形界面，处理工作和生活中的各类任务。针对 Blender、CAD 等专业软件和跨应用长程任务，混元了开展专项训练，大幅提升了专业任务的规划与执行能力。 模型和产品深度 Co-Design ，持续解锁真实生产力场景 得益于模型基础能力的全面提升，以及模型与产品的深度Co-Design，Hy4 解锁了更多真实生产力场景。 在办公场景，Hy4 全面提升了 WorkBuddy 的用户体验。评测结果显示，Hy4 平均效果得分 83.18 分，任务通过率 96.15%，12 个场景中 9 个排名国内第一。实际体验中，Hy4 在视觉创意、3D 场景构建和网页开发等任务上兼顾效果与效率。同时，针对WorkBuddy的人机双写特色功能，Hy4 在文档美观度、稳定性和可用性上均有明显提升。 伴随 Hy4 上下文学习能力的持续提升，模型能更好地理解多源复杂信息，成为用户的长期个人助理。借助 WorkBuddy 的企业微信套件，Hy4 可以连通聊天记录、邮件、文档、日程与会议等日常工作信息。多个项目并行推进时，Hy4 能理清各项目的推进历史与最新进展，识别待跟进的任务、需要准备的会议和临近截止的事项，及时提醒用户并协助安排优先级。此外，由 Hy4 支撑的团队协作功能将随新版“项目”在 Workbuddy 上线。 Hy4 重点强化了专业金融分析能力。基于真实金融场景构建了数千个工作空间和上万个高质量任务，让模型充分学习基金研究、证券分析、量化策略、资产配置、公司金融、财税会计等方向的专业流程与规范。混元联合腾讯金融科技 FiT 开展了 213 位专家的金融众测，其中 153 位来自外部的头部金融机构，且七成拥有五年以上工作经验。基于 579

Related stories

  • Google launches Gemini 4 Argon frontier model with API rollout

    Google DeepMind has released Gemini 4 Argon, the first frontier model in the Gemini 4 series, claiming it outperforms rivals like GPT-6 Astra and Claude Opus 5.5 across benchmarks in coding, cybersecurity, science, finance, and law. The model is available globally via the Gemini API, to Google AI Pro subscribers in the Gemini app and AI Mode, and in Antigravity for Ultra users. Launch pricing is $2 per million input tokens and $10 per million output tokens, with a 50% discount until November 30, after which prices double on December 1.

    8 reports · 6 sources1459 points · 1038 comments

  • Tencent unveils Buddy One cross-app AI collaboration system, Hunyuan Hy4 coming soon

    At the 2026 Tencent Global Digital Ecosystem Conference on October 29, Tencent launched Buddy One, a cross-app AI collaboration system for individual users. Buddy One uses WorkBuddy as its workbench, connects products like Tencent Docs, Tencent Meeting, Tencent Drive, and ima, and unifies accounts, points, and cross-product context retrieval via Buddy One ID. Tencent also announced upcoming upgrades to WorkBuddy Enterprise and the imminent release of Hunyuan Hy4 and Hy4 flash.

    2 reports · 1 sources

  • Google DeepMind unveils Gemini 4 Carbon, tops AI benchmark

    Google DeepMind announced Gemini 4 Carbon, its next frontier model, claiming state-of-the-art performance in cybersecurity and long-horizon tasks. The model scored 59 on the Artificial Analysis Intelligence Index v5.0, ranking first ahead of Claude Opus 5.5 (58) and GPT-6 Astra (57). It is initially available only to select cybersecurity defense companies through the Fairwind program, with pricing at $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 later.

    8 reports · 6 sources1013 points · 801 comments

  • Mistral releases open weights for 1T-parameter Le Chonk model

    Mistral AI has released the open weights for Mistral Large 4, also known as Le Chonk, a 1-trillion-parameter multimodal model trained in Europe. The company claims it is the best open-weight model for coding and rivals top open models like GLM-5.3 on the Artificial Analysis Coding Agent Index. The model is now available on the Hugging Face platform, and Mistral says its open nature gives organizations control over cyber defenses that closed models cannot offer.

    4 reports · 4 sources1709 points · 438 comments

News updates every hour. Subscribe in your RSS reader to get new stories automatically.

Subscribe