In the early morning of March 6, Alibaba released and opened source a new inference model Qwen QwQ-32B. Through large-scale reinforcement learning, the QwQ-32B has achieved a qualitative leap in mathematics, code and general capabilities, and the overall performance is comparable to DeepSeek-R1. At the same time, the QwQ-32B also significantly reduces the deployment cost, and can also be deployed locally on consumer graphics cards. At present, Ali has adopted the loose Apache2.0 protocol to open source the QwQ-32B model to the world, and everyone can download and commercialize it for free. At the same time, users will also be able to experience the latest Qwen QwQ-32B model for free through the Tongyi APP
Ali released and open source inference model Qwen QwQ-32B
In the early morning of March 6, Alibaba released and opened source a new inference model Qwen QwQ-32B. Through large-scale reinforcement learning, the QwQ-32B has achieved a qualitative leap in mathematics, code and general capabilities, and the overall performance is comparable to DeepSeek-R1. At the same time, the QwQ-32B also significantly reduces the deployment cost, and can also be deployed locally on consumer graphics cards. At present, Ali has adopted the loose Apache2.0 protocol to open source the QwQ-32B model to the world, and everyone can download and commercialize it for free. At the same time, users will also be able to experience the latest Qwen QwQ-32B model for free through the Tongyi APP
.