DeepSeek发布两款新模型

Core Insights - DeepSeek has released two official model versions: DeepSeek-V3.2 and DeepSeek-V3.2-Speciale [1][4] Group 1: Model Features - DeepSeek-V3.2 aims to balance reasoning capability and output length, making it suitable for everyday use. In benchmark tests, it reached the level of GPT-5, slightly below Gemini-3.0-Pro, while significantly reducing output length compared to Kimi-K2-Thinking, thus decreasing computational overhead and user wait time [3] - DeepSeek-V3.2-Speciale is designed to push the reasoning capabilities of open-source models to the extreme, exploring the boundaries of model capabilities. It is an enhanced version of DeepSeek-V3.2 with long-thinking capabilities and incorporates theorem-proving abilities from DeepSeek-Math-V2 [3] Group 2: Achievements - The V3.2-Speciale model has successfully won gold medals at several prestigious competitions, including IMO 2025, CMO 2025, ICPC World Finals 2025, and IOI 2025 [3] Group 3: Availability - The official web version, app, and API have all been updated to the official version of DeepSeek-V3.2. The Speciale version is currently available only as a temporary API service for community evaluation and research [4]