Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Tencent Hunyuan

AI overview

Sign in and the AI will write an overview from our coverage.

Headlines · 45

  1. Tencent Hunyuan reportedly open-sources Hy4; MetaX claims Day 0 GPU support

    36Kr, citing MetaX, reported that Tencent Hunyuan open-sourced Hy4 on November 20. MetaX also said its Xiyun C-series GPUs were the first to complete Day 0 adaptation for Hy4. The supplied report provides no independent verification of the adaptation claim.

    36氪快讯 · 🔥 4
  2. Tencent Hunyuan outlines larger and smaller Hy models

    Tencent Chief AI Scientist Yao Shunyu and members of the Hunyuan team held a 120-minute Reddit AMA about Hy4, Hy4 flash and the model roadmap. The team said Hy4 has substantial room for improvement, particularly through post-training, and that it plans to train larger models. It also said smaller Hy models and a Flash-Lite variant are on the roadmap, along with a future model targeting 24GB of graphics memory. The team stated that Hy3 and Hy4 demonstrate that well-executed smaller models can outperform larger models at lower cost.

    IT之家 AI · 🔥 16
  3. Tencent Hunyuan outlines Hy4 roadmap and architecture trade-offs

    Tencent’s Hunyuan team held a two-hour Reddit AMA discussing the Hy4 model family, its architecture, deployment plans, evaluation practices, safety restrictions and product roadmap. The team said the next Hy model will be significantly larger, while Hy4 mini and Flash-Lite were in development. It also acknowledged Hy4’s tendency to overthink and said recursive self-improvement remains at an early stage. The AMA followed Tencent’s opening of Hy4’s weights and release of Hy4 flash; reported cost and ranking figures came from Tencent disclosures or third-party rankings.

    钛媒体 · 🔥 9
  4. Chinese tech giants diverge on AI spending and revenue

    A 36Kr analysis of Alibaba, Baidu, Tencent and JD.com’s 2026 third-quarter results found different AI revenue and investment models. Alibaba reported RMB60.973 billion in AI cloud and computing-services revenue, up 53% year on year, while Baidu reported RMB13.2 billion in AI core-business revenue. The companies did not clearly isolate AI’s contribution to advertising, e-commerce or supply-chain growth. Investment also affected results: Alibaba’s AI labs and applications recorded a wider adjusted EBITA loss than a year earlier, while Tencent made RMB65.1 billion in capital-expenditure cash payments, including substantial advance purchases of computing capacity. JD.com reported that AI-related research and development would put short-term pressure on its margins.

    36氪 人工智能 · 🔥 0
  5. Tencent Hunyuan releases Hy4 model weights as open source

    Tencent Hunyuan announced that Hy4 model weights are available on GitHub, Hugging Face, ModelScope and GitCode for download and deployment. The reports say this completes the open sourcing of the Hunyuan LLM series, including Hy3 preview, Hy3, Hy4 preview, Hy4 and Hy4 flash. Hy4 uses a MoE architecture with 770B total parameters and 49B active parameters, with a context window exceeding 1 million tokens. Hy4 and Hy4 preview use the Apache 2.0 license. Tencent Hunyuan also published a Hy4 System Card covering cybersecurity and biosecurity evaluations.

    IT之家 AI · 🔥 0
  6. Tsinghua Yao Class alumni form a prominent global AI network

    A 36Kr article republished from Phoenix Weekly uses Tencent chief AI scientist Yao Shunyu’s November 18 lecture at Tsinghua’s Yao Class to examine the school’s role in AI talent. It identifies alumni in AI research and leadership roles at Tencent, DeepSeek, OpenAI, Meta and xAI, as well as in academia, infrastructure and applied technology. The report says claims that Yao Class alumni earn exceptional salaries, including claims about Yao Shunyu’s Tencent compensation and DeepSeek internships, have been debunked.

    36氪 人工智能 · 🔥 0
  7. Tencent’s Q3 AI spending pressures margins and cash flow

    TMTPost reports that Tencent’s AI-driven capital expenditure rose from RMB52.8 billion in Q2 to RMB65.1 billion in Q3, up 225% year over year, as infrastructure expanded for model upgrades and WorkBuddy inference. Tencent’s operating margin fell from 33% to 29%, while free cash flow swung from negative RMB13.8 billion in Q2 to negative RMB69.1 billion in Q3, though it returned to positive RMB16.9 billion when compute-related prepayments were excluded.

    钛媒体 · 🔥 0
  8. Tencent opens Hy4 flash weights and adapts it to MetaX GPUs

    Tencent Hunyuan has formally opened the model weights for Hy4 flash, according to 36Kr. MetaX said its Xiyun C-series GPUs completed Day 0 end-to-end adaptation for Hy4 flash, with inference validation on the same day. MetaX also claimed the setup can support high-throughput, low-cost, scaled Agent workloads, but no performance figures were provided.

    36氪快讯 · 🔥 0
  9. Tencent releases and open-sources Hy4 flash model family

    Tencent Hunyuan released Hy4 flash, a natively multimodal model with a 1M-token context window, and made the standard, FP8 and Base variants available for download on major model platforms. Tencent describes it as its most cost-efficient Hy model, with 12B active parameters per token and substantially lower API pricing than Hy4. Independent office-workflow testing found Hy4 flash produced usable documents and web pages with relatively low usage, while most benchmark and cost comparisons remain vendor-defined.

    Hacker News 首页(100+ 分) · 🔥 0
  10. Tencent Hunyuan’s Hy4 Flash launches on OpenRouter with 50% discount

    Tencent Hunyuan announced that Hy4 Flash is now available through OpenRouter. Under a limited-time 50% discount, the listed rates are $0.035 per 1 million input tokens, $0.1405 per 1 million output tokens, and $0.0015 per 1 million cached-read tokens. The duration of the discount was not specified. Tencent Hunyuan also cited regular API prices of $0.070 for input, $0.281 for output, and $0.003 for cached reads per 1 million tokens. The company claimed that Hy4 Flash matched Claude Haiku 5.5 (high) in human blind tests on daily work while reducing task cost by 18 times.

    腾讯混元 · 🔥 0
  11. Tencent Hunyuan launches Hy4 Flash for low-cost AI workloads

    Tencent Hunyuan released Hy4 Flash, a multimodal MoE model with 224B total parameters, 12B active parameters per token and a 1M-token context window. Tencent says mixed attention, LatentMoE, Engram, quantization and other inference optimizations reduce memory use and enable decoding above 150 tokens per second, while internal WorkIndex testing puts its per-task cost at one-twentieth of Hy4. The model is live across Tencent Cloud TokenHub, WorkBuddy, Yuanbao, WeCom and other products, with an open-source release planned; independent tests by ifanr and GeekPark support its speed-and-cost positioning but also found less refined output in some creative tasks.

    腾讯混元 · 🔥 0
  12. Tencent calls for AI-era business reinvention and flatter teams

    Tencent Culture published field notes from the company’s 2026 executive LinkTime on Nov. 16. The notes called for AI-native businesses, models and products to move forward, while every core business team should reinvent itself for the AI era. Management proposed flatter organizations, small teams as operating units and splitting core-business staff between maintaining existing operations and pursuing new initiatives. Tencent described an AI product stack comprising foundation models, Xiaowei for personal assistance, Yuanbao for information and decision support, and WorkBuddy as an AI-native workplace product.

    IT之家 AI · 🔥 3
  13. Tencent Hunyuan launches Hy Realtime voice model

    Tencent Hunyuan released Hy Realtime, a real-time voice model with Agentic tool calling, singing, contextual intent recognition, and optional web search. The company says it is already used in Yuanbao voice calls and Tencent customer service, with a Tencent Cloud API planned; users can try the "初语" voice in the latest Yuanbao app. Tencent Hunyuan also says future development will focus on duplex interaction, personalized long-term memory, and foreground-background Agent interaction.

    IT之家 AI · 🔥 0
  14. Tencent Hunyuan Hy4 tops CWE-Bench-AA and ranks third on Cyber Index

    Artificial Analysis reports that Tencent Hunyuan Hy4 scored 58% on its Cyber Index, ranking third behind Kimi K3.1 Pro and GPT-6 Sol (Daybreak Blue, max). Hy4 placed first on CWE-Bench-AA with 76%, leading DeepSeek V4.1 Flash 1017 (max) by six points in an implementation of Collinear AI's CWE-bench v1. Artificial Analysis also reported API pricing of $0.834 per million input tokens and $2.501 per million output tokens, with a cost of $0.92 per task on its capability-cost frontier.

    ArtificialAnlys (X) · 🔥 0
  15. Tencent Cloud WAND targets rework costs in AI content production

    InfoQ Chinese reported that Tencent Cloud WAND combines 3D white-model constraints, scene-specific model tuning, local video repair, and Agent-based workflow orchestration to reduce repeated generation in AI video and image production. Tencent Cloud said WAND reduced customers' overall content-production costs by about 20%, while named customers reported higher usable-output rates and additional cost or efficiency gains. These performance and savings figures were supplied by Tencent Cloud and the interviewed customers and were not independently verified.

    InfoQ 中文 AI&大模型 · 🔥 0
  16. ByteDance and Tencent AI leaders reportedly depart amid team restructuring

    36Kr, citing Z Finance, reported that ByteDance Seed core lead Zhou Chang left to found a Physical AI startup backed by major dollar funds. Citing multiple sources, 36Kr also reported that Tencent Hunyuan chief research scientist and multimodal reinforcement-learning lead Pang Tianyu departed after less than nine months. The article connects these departures to broader organizational shifts in Chinese multimodal AI teams, but its claim that multimodal generation is losing priority remains analysis rather than established fact.

    36氪 人工智能 · 🔥 0
  17. Compute Costs Reshape China’s Independent Model Companies

    A Tech商业 article republished by 36氪 argues that scarce compute is pressuring Kimi, Zhipu, MiniMax and StepFun to adopt different strategies: securing GPUs through capital, building infrastructure, shifting toward higher-margin business revenue, and exploring mobile deployment. It frames these responses as four ways independent model companies are paying an increasingly rigid “compute rent,” while arguing that large technology companies’ infrastructure and distribution advantages are weakening the case for four comparable independent contenders.

    36氪 人工智能 · 🔥 0
  18. Tencent Hunyuan Hy4 launches with visual and computer-use capabilities

    Tencent unveiled Hunyuan Hy4 at the Wuzhen Summit, adding visual understanding and computer-operation capabilities. According to the report, the model can interpret charts and documents and operate graphical interfaces and command-line tools. The report inconsistently describes the release as fully open source while also saying the model weights will be released soon.

    掘金 · 人工智能 · 🔥 0
  19. Tencent Hunyuan Hy4 Tops Chinese Models in Agent Arena

    Arena reports that Tencent Hunyuan’s Hy4 ranked first among Chinese models and 14th overall in Agent Arena, based on thousands of real-world, long-horizon agent sessions. It placed on the benchmark’s Pareto frontier, with a reported median cost of $0.55–$0.57 per task, and improved by 3.6 percentage points over Hy4 preview. Hy4 remains proprietary, while Arena’s claim that it would tie Kimi K3.1 Pro for first among open models is conditional on a future weight release.

    arena (X) · 🔥 0
  20. Tencent Hunyuan Expands Hy4 Capacity for WorkBuddy

    Tencent Hunyuan and WorkBuddy said Hy4 reached many users' limits on its first day in WorkBuddy, prompting an urgent tenfold expansion of the Hy4 inference cluster and an increase in free usage allowances. IT之家 AI and 36氪快讯 reported the vendor announcement. Tencent said limits could still occur during peak periods, while Hy3's free use in WorkBuddy was extended to November 15, 2026, at 23:59.

    IT之家 AI · 🔥 0
  21. Tencent Releases Hy4 Open-Weight Foundation Model

    Bloomberg reported that Tencent Holdings Ltd. released the Hy4 open-weight foundation model on Wednesday. Tencent said Hy4 joins the top tier of Chinese open-weight AI systems for the first time, a vendor-attributed claim. The company also plans a Flash version.

    Bloomberg Technology · 🔥 0
  22. Tencent releases Hy4 with vision and computer use

    Tencent Hunyuan released the 770B-parameter Hy4 model, expanding pre-training and post-training over Hy4 preview while adding vision perception, computer use, improved long-context understanding and stronger long-horizon task performance. Tencent says Hy4 is available through Tencent Cloud, OpenRouter and products including WorkBuddy and CodeBuddy. Model weights and a system card are still planned, so the release does not yet constitute an open-weight publication. Artificial Analysis scored Hy4 at 35 on its Intelligence Index, placing it second among Chinese lab models but behind Kimi K3.1 Pro, while Tencent acknowledged a remaining gap with leading proprietary models.

    Hacker News 首页(100+ 分) · 🔥 0
  23. China’s AI Cloud Market Shifts Toward Full-Stack Competition

    A commentary from 巨潮WAVE argues that AI cloud platforms are becoming the central infrastructure for model training, inference, and agent development, as competition shifts from scale and price toward integrated capabilities. It compares Alibaba Cloud, Tencent Cloud, China Telecom’s 天翼云, and Baidu AI Cloud across chips, models, ecosystems, infrastructure, and capital investment. 虎嗅 AI and 钛媒体 published the same authored article, so the two reports do not independently corroborate its market estimates or vendor-related claims.

    钛媒体 · 🔥 0
  24. Jinwen AI Launches for Shanxi Culture With Seven Cultural Agents

    Jinwen AI, a vertical model focused on Shanxi culture, has launched alongside its digital human Jin Dou. The system uses an LLM+RAG+Agent architecture, integrates Tencent Hunyuan and DeepSeek technologies, and draws on reviewed books from eight publishers affiliated with Shanxi Publishing Media Group. Its seven cultural agents cover topics including historic architecture, Jin merchants, family origins, cultural and creative design, prominent figures, and military history.

    AIbase AI新闻 · 🔥 0
  25. Ten Governance Conflicts Facing Leading Foundation-Model Companies

    A commentary from the WeChat publication Enterprise Management Magazine, written by Xu Weina and Xiaoshan Zhang, examines ten structural conflicts facing leading foundation-model companies. The article groups OpenAI, Anthropic, DeepSeek and Moonshot AI as independent AI startups, while Google DeepMind, Meta FAIR and ByteDance Seed represent AI businesses embedded within larger technology groups. It discusses tensions involving public missions and capital, research autonomy and corporate control, decentralized versus centralized decisions, research and commercialization, technical and customer priorities, long-term research and short-term delivery, open versus closed models, technical and commercial thinking, weak versus strong performance evaluation, and open debate versus coordinated action. It proposes approaches including segregated resources, protected long-term research, differentiated management and layered decision rights.

    虎嗅 AI · 🔥 0
  26. Tencent Hunyuan launches Hyra 2.0 for reusable AI research tools

    Tencent Hunyuan introduced Hyra 2.0, an AI research system designed to move beyond solving individual problems toward building reusable models, workflows, and software tools. According to IT之家 AI and AIbase AI新闻, its three-layer architecture assigns coordination to Overall, research direction to Mission, and experiments and analysis to Worker, while separating project state, agent memory, and a shared Books knowledge base. AIbase AI新闻 also reported claimed results from HyInv, HyPLM-Flash, robot-hand design, and Xiangshan processor optimization, but these figures have not been independently verified.

    腾讯混元 · 🔥 0
  27. Tencent Hunyuan Reinforcement-Learning Lead Reportedly Leaves

    It was reported that Pang Tianyu, Tencent Hunyuan’s reinforcement-learning lead, has left the company.

    DoNews · 🔥 0
  28. Tencent Hunyuan Restructures Foundation-Model Organization

    A 36Kr article citing workplace social networks and AI-industry discussions reports that Tencent Hunyuan has undergone organizational restructuring marked by department consolidation, shorter reporting lines and localized personnel changes following Yao Shunyu’s arrival. The report says Tencent has not characterized the changes as layoffs or a large-scale workforce reduction. At Tencent’s May 2026 shareholder meeting, President Martin Lau said the company had no major layoff plan. The report also says Tencent merged its large-language and multimodal-model departments into a foundation-model department in July 2026, with Yao overseeing language, multimodal, reinforcement-learning and agent work.

    36氪 人工智能 · 🔥 0
  29. WeChat Pay’s Hunyuan-Based Robot Debuts in a 7-ELEVEN Store

    A WeChat Pay smart business robot made its first store deployment at the opening of the 2,000th Guangdong 7-ELEVEN location in Shenzhen, according to IT之家. The robot can greet shoppers, recommend products and promotional bundles, take orders and help with self-checkout, while Tencent says its Hunyuan foundation provides multi-turn dialogue and Agent capabilities for sales, payments and operations. It also uses Tencent Tianlai inside audio technology, which the vendor says helps it recognize speech in noisy stores.

    IT之家 AI · 🔥 0
  30. Tencent upgrades WorkBuddy Enterprise with security, observability and managed agents

    Tencent announced four core upgrades for WorkBuddy Enterprise Edition: an open platform, an enterprise security solution, an observability center and WorkBuddy Managed Agents (WMA), according to Zhidx and InfoQ. The open platform connects Tencent workplace products with third-party IM platforms, enterprise systems and supported models. WMA moves task execution into the cloud and can be used through the WorkBuddy management console or Tencent Cloud console, with API, SDK, A2A and common IM-platform integrations. Tencent also released a WorkBuddy Enterprise security white paper covering identity, permissions, execution protection, data security and auditing.

    InfoQ 中文 AI&大模型 · 🔥 0

Next →

Experience and discussion from the community

Share my Tencent Hunyuan experienceAsk about Tencent Hunyuan

Nobody has shared their experience with Tencent Hunyuan yet.