DeepSeek V3.2 official version released: Enhanced Agent capabilities, integrated thinking and reasoning.

Golden Finance reported that today we are simultaneously releasing two official version models: DeepSeek-V3.2 and DeepSeek-V3.2-Speciale. DeepSeek-V3.2 is our first model that integrates thinking into tool usage, and it supports both thinking mode and non-thinking mode for tool invocation. We have proposed a large-scale agent training data synthesis method, constructing a large number of “hard to answer, easy to verify” reinforcement learning tasks (over 1800 environments, over 85,000 complex instructions), significantly improving the model's generalization ability. (DeepSeek)