Tencent has made its Hy3 large language model available globally through WorkBuddy (its AI agent workspace), Design Miora (creative studio), and Cloud TokenHub, with free access through August 31, 2026 (Pacific Time). Hy3—a 295-billion-parameter model with 21 billion active parameters and 256K context window—is available under the Apache 2.0 license on Hugging Face, ModelScope, and OpenRouter at $0.1288 per million input tokens and $0.5336 per million output tokens.
In internal Tencent workplace evaluations, Hy3 delivered a task success rate exceeding 90% when used in WorkBuddy while reducing average task completion time by 34% compared to the previous Hy model iteration. Hy3 ranked first on OpenRouter's global LLM usage leaderboard within one week of launch, recording 68 times more API calls than its predecessor. The model is integrated into Tencent's product ecosystem and available through partner platforms including Cafe24 (South Korea e-commerce) and Metelix (Japan enterprise AI).
For teams evaluating open-weight reasoning models and agent frameworks, Hy3's 90%+ task success in workplace scenarios and 34% efficiency gain offer a compelling efficiency signal at competitive token pricing. WorkBuddy's free tier through end of August provides a low-friction trial window for multi-agent orchestration, with subsequent per-token billing supporting cost-sensitive inference deployments.