On June 16, Tencent Cloud revealed that, as per its adjustment strategies, the DeepSeek-V3.2 model, which was initially offered on both the Knowledge Engine Atomic Capability Platform and the Large Model Service Platform TokenHub, will undergo an official discontinuation at the stroke of midnight (00:00) Beijing Time on July 16, 2026. Following this date, users will no longer have access to this model. Tencent Cloud strongly advises users to transition to the more advanced DeepSeek-V4 series models. Should users fail to accomplish this migration, the system will automatically default to the latest DeepSeek model to maintain uninterrupted service. Furthermore, commencing at 00:00 on June 3, Tencent Cloud's Intelligent Agent Development Platform has implemented price reductions for the DeepSeek-V4 series models. Notably, the input and output costs for inference operations of the DeepSeek-V4-Pro model have been slashed by 75%, while cache hit prices have seen a reduction of up to 97.5%. Similarly, the cache hit prices for the DeepSeek-V4-Flash model have been decreased by 90%.