Tencent Open-Sources Hy4 Preview, a 770B-Parameter MoE AI Model
Tencent released Hy4 Preview, an open-source Mixture-of-Experts AI model with 770 billion total parameters (49 billion active), a 1-million-token context window, and agentic research capabilities. Released under the Apache 2.0 license, the model excels at multi-step software engineering tasks and autonomous coordination. It is available for free for two weeks on the WorkBuddy_AI platform, marking a major advancement in Chinese open-source large language models.
Editorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page itself is projected from evidence records.
- Current automated evidence projection
Cross-source coverage
Wire timeline
Tencent Hunyuan reportedly gives entire team week off after Hy4 AI model exceeds expectations
A social media post claims that Tencent Hunyuan rewarded its entire team with a one-week vacation after the performance of its Hy4 AI model exceeded internal expectations. The news has not been officially confirmed by Tencent. However, Hy4 has been performing impressively in recent days. On August 28, Tencent released and open-sourced the Hy4 preview, which features a total of 770 billion parameters and activated parameters. The report highlights the team's apparent celebration of the model's success, though verification is pending.
Tencent releases Hy4 preview with 770B parameters and 1M token context window
Tencent's Hunyuan engineering team has released the Hy4 preview, a large language model with 770 billion total parameters and 49 billion active parameters. The model features a 1 million token context window designed for massive codebase retention and sustained long-horizon execution. It also supports advanced multi-source tool use. The model is specifically built to handle long-horizon software engineering and multi-step agentic workflows without losing context. Benchmarks reportedly back up the model's scale and performance claims.
Tencent Releases Hy4 Preview, a 770B MoE Model for Agentic Tasks and Software Engineering
Tencent has launched the preview of Hy4, a 770-billion parameter Mixture of Experts (MoE) model specifically designed for multi-step agentic tasks and software engineering. The model is available for free for two weeks within the WorkBuddy_AI platform. The announcement was made via a social media post, which also included a demonstration of the model's capabilities in building a standalone 3D Voxel Zen Garden, showcasing its proficiency in long-horizon frontend development tasks.
Show 4 older updatesHide older updates
Tencent Hunyuan Open-Sources Hy4 Preview with 770B Parameters
Tencent Hunyuan officially open-sourced its next-generation flagship AI model, Hy4 Preview, today. The model features 770 billion total parameters with 49 billion activated per token, a native 1 million token context length, and is available in both original and FP8 weights under the Apache 2.0 License. The architecture is notably aggressive, with 77 out of 78 layers using Mixture of Experts (MoE), each containing 256 routing experts plus one shared expert, activating the top 8 experts per token. This release marks a significant step in open-source large language model development, offering high performance and long-context capabilities.
Tencent Hunyuan Releases Hy4 Preview, a 770B-Parameter Open-Weight MoE Model
ChrisGPT, a news intelligence source, reports that Tencent's Hunyuan team has released a preview of their new AI model, Hy4. The model is a massive Mixture of Experts (MoE) architecture with 770 billion total parameters and 49 billion active parameters per token. It features a 1 million token context window. The model weights have been released under the permissive Apache 2.0 license, making it openly available. The post includes a link to further details, though the caveat is not elaborated in the text. This release marks a significant development in open-weight AI models from a major Chinese technology company.
Tencent Releases Hy4 Preview: 770B-Parameter Open-Source Model with Agentic Research Capability
Tencent has released Hy4 Preview, a 770-billion-parameter open-source AI model with 49 billion active parameters and a 1-million-token context window. The model's standout feature is its agentic research capability, which allows it to coordinate multiple Codex sessions in parallel and evaluate their results. This release marks a significant advancement in open-source large language models, emphasizing autonomous research and multi-agent coordination.
Tencent Releases Open-Source AI Model Hy4 with 770B Parameters
Tencent has released a new open-source AI model called Hy4, as announced on X by the account ai_for_success. The model features 770 billion total parameters, with 49 billion active parameters, and supports a context length of 1 million tokens. This release adds to the growing list of Chinese open-source AI models, highlighting Tencent's entry into the competitive large language model space. The post includes a link to further details, but the core announcement is clear and newsworthy for the AI and tech industry.