today, we are releasing two official versions of models: DeepSeek-V3.2 and DeepSeek-V3.2-Speciale. DeepSeek-V3.2 is our first model that integrates reasoning into tool usage and supports both reasoning mode and non-reasoning mode 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 (1800+ environments, 85,000+ complex instructions), significantly improving the model's generalization ability. (DeepSeek)