On December 5, the latest Tencent Hybrid language models Tencent HY 2.0 Think and Tencent HY 2.0 Instruct were officially released. HY 2.0 uses a hybrid expert architecture with 406B total parameters and 32B activation parameters. It supports a 256K context window. Its reasoning ability and efficiency rank among the top in the country, and has outstanding performance in practical scenarios such as text creation and complex instruction compliance. Compared with the previous version of the model, HY 2.0 Think has significantly improved pre-training data and reinforcement learning strategies. The overall performance in complex reasoning scenarios such as mathematics, science, code, and instruction following has been steadily ranked first in the country, and generality has been greatly improved.

On December 5, the latest Tencent Hybrid language models Tencent HY 2.0 Think and Tencent HY 2.0 Instruct were officially released. HY 2.0 uses a hybrid expert architecture with 406B total parameters and 32B activation parameters. It supports a 256K context window. Its reasoning ability and efficiency rank among the top in the country, and has outstanding performance in practical scenarios such as text creation and complex instruction compliance. Compared with the previous version of the model, HY 2.0 Think has significantly improved pre-training data and reinforcement learning strategies. The overall performance in complex reasoning scenarios such as mathematics, science, code, and instruction following has been steadily ranked first in the country, and generality has been greatly improved.