Tencent opens Hy3 large language model to global users across products and cloud

Global rollout through products and cloud services

Tencent announced broader international access to its Hy3 large language model on August 5, 2026, extending the system beyond its home market through the WorkBuddy agent workspace, the Tencent Design Miora design studio, and the Tencent Cloud TokenHub Model-as-a-Service platform. The model is offered free of charge on WorkBuddy to users worldwide until August 31, 2026 (Pacific Time), while developers can also connect via API and integrate it with coding extensions, third-party platforms, and other tools. Tencent described the move as part of a rollout that broadens access "for developers, users and enterprises, offering multiple ways to experience and integrate the model." The Hy3 release builds on the official Hy3 launch on July 6, 2026, and follows the rebranding of Tencent's Hunyuan team as Tencent Hy.

Model architecture and specifications

Hy3 is a hybrid fast-and-slow-thinking model built on a Mixture-of-Experts architecture, with 295 billion total parameters and 21 billion active parameters, and supports a context length of up to 256K tokens. Tencent says the system performs comparably to flagship models with two to five times as many parameters across reasoning, instruction understanding, in-context learning, code generation, and agent capabilities. The model weights are available on Hugging Face and ModelScope under the commercially permissive Apache 2.0 license. In internal evaluations conducted within Tencent's workplace environment, Hy3 delivered a task success rate exceeding 90% in WorkBuddy while reducing average task completion time by 34% compared with the previous Hy iteration, according to the company.

Measured usage and rankings

Tencent said Hy3 recorded more than 68 times as many API calls as its previous-generation model and ranked first globally on OpenRouter's LLM usage leaderboard within one week of launch. Since the Hy3 preview released on April 23, 2026, average daily token consumption rose twenty-fold and the number of WorkBuddy users actively selecting Hy3 preview grew sixfold, the company added. Hy3 is also distributed through developer platforms including Hermes, Kilo, Cline, OpenClaw, OpenCode, and Cherry Studio, and is integrated across Tencent's own product surface, including the Yuanbao assistant, the ima knowledge-base tool, the Marvis OS-level agent, and customer-service assistants on Weixin Official Accounts.

Pricing and access channels

Hy3 API access on OpenRouter is listed starting at US$0.1288 per million input tokens and US$0.5336 per million output tokens, providing a per-token rate that Tencent describes as a "cost-efficient foundation" for deploying AI applications. WorkBuddy's free tier ends after August 31, 2026 (Pacific Time). Tencent said further Hy3 integrations are planned across its AI-native product portfolio, alongside regional partnerships with South Korea-based Cafe24 and Japan-based Metelix to distribute the model through their AI gateway and platform services.

What remains uncertain

The sources do not specify how Hy3's ranking or API call figures are measured against comparable baselines, nor do they detail the post-August-31 pricing on WorkBuddy. The extent of adoption among international enterprises beyond the announced Cafe24 and Metelix partnerships is not quantified in the available reporting.