HY4 Preview
AI overview
Sign in and the AI will write an overview from our coverage.
Headlines · 8
- Tencent Hunyuan Hy4 model weights are now open source
Tencent Hunyuan opened the Hy4 model weights on GitHub, Hugging Face, ModelScope and GitCode, allowing developers to download and deploy them. According to IT之家, Hy4 uses a MoE architecture with 770B total parameters, 49B active parameters per token and a context window exceeding 1 million tokens; cnBeta also reports text, image and video understanding. Tencent released an accompanying System Card, while Hy4 and Hy4 preview use the Apache 2.0 license.
IT之家 AI · 🔥 14 - WorkBuddy Tops 10 Million Daily Users and 100 Trillion Daily Tokens
Tencent disclosed in its 2026 third-quarter financial results that WorkBuddy had more than 10 million daily active users and exceeded 100 trillion Token calls per day as of early November. The company reported that the platform had completed over 100 major updates since March, while its open platform covered 55 professional scenarios and more than 30 vertical sectors. Tencent also said WorkBuddy Enterprise had reached over 50 industries and half of China’s top 500 enterprises.
IT之家 AI · 🔥 0 - Tencent Cloud outlines enterprise AI MaaS stack
At Tencent Cloud’s MaaS session, the company presented Hunyuan, TI-ONE, TokenHub and partner model practices as a stack spanning model development, deployment and business applications. It also described Hunyuan’s internal evaluation system, TI-ONE inference optimization on domestic accelerators and TokenHub’s launch of Xiaomi’s MiMo-V2.6 series. Tencent executives said model capabilities were improving, per-million-token call costs had fallen from yuan to jiao or fen, and agents were moving beyond chat interfaces toward planning and execution. The reported fivefold single-card throughput gain for Hy4 preview was measured in a specified domestic-accelerator test environment.
InfoQ 中文 AI&大模型 · 🔥 0 - Mini Program Growth Plan adds DeepSeek-v4.1-flash and Hy4 Preview to token allowance
According to 掘金·人工智能, the AI resource package included in WeChat's Mini Program Growth Plan now supports two newly added deductible models, hy4-preview and deepseek-v4.1-flash, bringing the 1-billion-token text-generation allowance to three supported text models including hy3. The report says none of the three need to be enabled in the cloud-development console, and token usage from mini-program clients, the cloud-development server or the console is deducted automatically from the gifted allowance without extra configuration. It also notes that the Hy-image-v3.5-preview image model and the newly released Hy4 are still being integrated, with a separate notice to follow.
掘金 · 人工智能 · 🔥 0 - 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 - 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 - GPT-6 Luna and GPT-6.1 Sol Report Six Agent Arena Frontier Positions
Arena reported that GPT-6 Luna and GPT-6.1 Sol occupy six of the 11 positions on the Agent Arena Pareto frontier across different reasoning efforts. It said Luna holds three of four frontier positions below $0.10 per task, while GPT-6.1 Sol occupies positions between $0.22 and $0.65 per task. Adding GPT-6 Astra at Max reasoning gives OpenAI seven of the 11 positions. The report was based on a single Arena post.
arena (X) · 🔥 0 - Tencent reportedly tests “Dongleya” AI learning community
IT之家 reported, citing user feedback, that Tencent is beta-testing “懂了鸭,” an AI interactive-content community for children and teenagers. The website and WeChat mini program reportedly offer limited testing to parents with children and frontline teachers. Its built-in AI agents can reportedly turn natural-language prompts into 3D demonstrations, dynamic lessons, interactive picture books, lightweight games and memory cards without requiring programming skills. Users can browse community works and adapt others’ creations. IT之家 also said the model list includes DeepSeek-V4.1-Flash, DeepSeek V4 Pro, DeepSeek V4 Flash, Hy3, Hy4 preview, MiniMax M3, Kimi K3 and GLM-5.3. The supplied reports do not include direct confirmation from Tencent, and they describe the test’s access arrangements differently.
DoNews · 🔥 0
Experience and discussion from the community
Share my HY4 Preview experienceAsk about HY4 Preview
Nobody has shared their experience with HY4 Preview yet.