Tencent has released its Hy3 AI model, which boasts 295 billion parameters and can handle context lengths of up to 256,000 tokens. Despite having only 21 billion active parameters, Hy3 allegedly matches the performance of models two to five times its size. In testing, Hy3 scored higher than GLM-5.1 and showed a significant reduction in hallucination rate. The model is now available under an open-source license, and Tencent has already integrated it into several of its products. Why it matters: Hy3’s impressive performance could lead to more efficient and scalable AI models, making it a significant development in the broader AI landscape.