DeepSeek founder Liang Wenfeng recently clearly stated in an internal communication that the highly anticipated next-generation flagship large model DeepSeek V4 will be officially released in late April 2026.

image.png

Layered Modes Emerge: The Battle Between Speed and Expertise

Before the official launch of V4, DeepSeek has made significant adjustments to its product side, introducing for the first time a layered mode design:

  • Fast Mode: Focuses on daily conversations and instant responses, supports text recognition in images and files, with a focus on "speed."

  • Expert Mode: Designed for complex logic and deep thinking, with stronger intelligent search capabilities. However, this mode currently does not support file uploads and multimodal functions, and may require waiting during peak hours.

Advancing Amidst Turmoil: Performance Improvements and Service Abnormalities Coexist

Although the release of the new model is approaching, DeepSeek's recent performance can be described as "painful yet joyful." On one hand, netizens have noticed significant improvements in its logical processing and coding capabilities; on the other hand, the platform has experienced large-scale service abnormalities for three consecutive days, including a crash lasting up to 12 hours. This has been viewed by the industry as a "pain period" during the transition between old and new models.

Notably, the release date of DeepSeek V4 is highly competitive. Previous reports have indicated that Yao Shunyu's team at Tencent will also release a new Hunyuan large model next month. As late April approaches, these two major domestic AI giants will engage in a direct competition in base model capabilities.