DeepSeek: Currently testing a new long text model structure, API services remain unchanged

AASTOCKS
2026.02.15 23:32

DeepSeek's official assistant stated in the developer group that the DeepSeek website and APP are testing a new long text model structure, supporting 1M context, while the API service remains unchanged, still at V3.2, only supporting 128K context.

This test is part of a gray test that started on February 11. The new model's context window has significantly improved compared to the previous 128K version, capable of processing text equivalent to 500 pages of A4 documents in a single instance, covering scenarios such as long document analysis and cross-chapter reasoning.

However, some users have reported changes in the interaction style of the new version. Industry speculation suggests that this version may be DeepSeek-V4-Lite, sacrificing some quality for speed, serving as a stress test for the official version

Source: AASTOCKS The copyright of this article belongs to the original author/organization.
The views expressed herein are solely those of the author and do not reflect the stance of the platform. The content is intended for investment reference purposes only and shall not be considered as investment advice. Please contact us if you have any questions or suggestions regarding the content services provided by the platform.
© 2026 Longbridge｜Disclaimer