The Tencent Hy3 artificial intelligence model is now accessible to global developers, users, and enterprises following its broader international rollout.
Users and enterprises can access the model through WorkBuddy, where it is available free of charge until August 31, 2026. In addition, the model is integrated into Tencent Design Miora and Tencent Cloud TokenHub, with further integrations planned across the company’s cloud portfolio. Developers can also connect to the model via API and integrate it with developer environments, coding extensions, and third-party software platforms.
Technical Specifications of Tencent Hy3
Built on a hybrid Mixture-of-Experts architecture, the Tencent Hy3 model features 295 billion total parameters and 21 billion active parameters. Specifically, it supports a context length of up to 256K tokens, performing tasks across reasoning, code generation, and instruction understanding.
According to internal evaluations, the model achieved a task success rate exceeding 90% when deployed within the WorkBuddy workspace. This represents a 34% reduction in average task completion time compared to the previous iteration of the model.
Integration and Platform Availability
The model is available on open-source communities including Hugging Face and ModelScope under the commercially permissive Apache 2.0 license. Meanwhile, developers can access the API through OpenRouter, with pricing starting at $0.1288 per million input tokens and $0.5336 per million output tokens. This pricing structure provides a cost-efficient foundation for deploying real-world applications.
The model has also expanded to third-party developer platforms such as Hermes, Kilo, Cline, OpenClaw, OpenCode, and Cherry Studio. Within one week of launch, it ranked first on OpenRouter’s global usage leaderboard. Notably, it recorded more than 68 times as many API calls as its previous-generation model.
Enterprise Deployment and Infrastructure
Poshu Yeung, Senior Vice President of Tencent Cloud, stated that the model helps organizations move beyond experimentation to practical deployments.
“By connecting advanced models with real-world scenarios, enterprise-grade engineering, flexible model orchestration and secure infrastructure, we aim to give enterprises the control and reliability they need to move from pilots to scaled deployments.”
Poshu Yeung, Senior Vice President of Tencent Cloud and Head of Tencent Cloud International
Global Cloud Infrastructure Support
Tencent supports these international deployments through its global cloud infrastructure, which spans 66 Availability Zones across 23 regions. This network is further supported by more than 3,200 acceleration nodes worldwide.
The company plans to continue expanding its AI-native product portfolio, integrating Tencent Hy3 into more enterprise workflows. Furthermore, this infrastructure ensures reliable access for global businesses scaling their digital operations.