Tencent has released the preview version of its Hunyuan Hy4 large language model, built on a mixture-of-experts (MoE) architecture with 770B total parameters, 49B activated at a time, and a 1-million-token context window. The new generation is more than twice the size of Hy3 — which had 295B total parameters, 21B active parameters and 256K context — and its long-context capacity has grown roughly four-fold.
Rather than lean only on general benchmarks, Tencent is steering Hy4 toward concrete tasks in coding, office work, gaming and scientific research. The model can build a Three.js 3D website from scratch, craft a playable game demo in Unity, or work through 72 financial documents at once to catch duplicate reimbursements, over-limit expenses and budget anomalies.
Hy4 was trained and iterated along with Tencent products including WorkBuddy and CodeBuddy, and it has begun contributing to its own development — helping optimize training methods, data strategy, evaluation systems and underlying operators before running experiments and adjusting its approach based on the results. The preview is already integrated into WorkBuddy, CodeBuddy, Yuanbao, ima, Tencent Cloud TokenHub and OpenRouter, with Tencent having flagged earlier in its earnings report that Hy4 would surpass Hy3 and that real product feedback would be a key source for training.
Tencent has released the preview of Hunyuan Hy4, a mixture-of-experts model. It carries 770B total parameters, activates 49B at a time, and supports a 1-million-token context window.
Doubling the scale, quadrupling the context
Hy4 follows Hy3, which packed 295B total parameters, 21B active parameters and a 256K context. This generation more than doubles that size, and its long-context ceiling is roughly four times higher. Tencent is not leaning only on benchmark scores this time. The company says it sharpened the model for real tasks in coding, office work, gaming and scientific research.
Three examples: a 3D site, a Unity demo, a finance sweep
Tencent spelled out a few use cases. Hy4 can build a Three.js 3D website from scratch, create a playable game demo inside Unity, and process 72 financial documents in one pass, flagging duplicate reimbursements, over-limit expenses and budget anomalies along the way. Hy4 was also trained and iterated together with Tencent products such as WorkBuddy and CodeBuddy.
A model helping to build the next version of itself
The more striking detail is that Hy4 has started contributing to its own development. According to Tencent, the model helps optimize training methods, data strategies, evaluation systems and underlying operators. It proposes plans, runs experiments, then adjusts based on the results. The code, logs and feedback generated by those experiments flow into the next round of development — an initial self-improvement loop that Tencent says is already taking shape.
Where the Hy4 preview is available now
The preview has been integrated with WorkBuddy, CodeBuddy, Yuanbao, ima, Tencent Cloud TokenHub and OpenRouter. Tencent had already said in its earnings report that Hy4 would be bigger than Hy3, and that real product feedback would be an important source for model training.
This article was originally published by Bit.Fan. For more cryptocurrency news and market insights, visit www.bit.fan. Disclaimer:
The market information, project data, and third-party content displayed on this platform are for industry information sharing only and do not constitute any form of investment advice or return commitment.
Cryptocurrency trading carries high risks. Users should fully assess their risk tolerance and make independent decisions. All profits, losses, and legal responsibilities are borne by the users themselves.
