DeepSeek updated V4 Pro on August 13 across its app, website, and API. The update focuses on agent workflows and compatibility with coding tools, including Codex.
The original release note also described three reasoning settings for V4 Flash and Pro: low, high, and max. They let you choose an effort level appropriate to the task.
The pricing chart uses peak and off-peak rates, with the announced changes taking effect on August 17, Beijing time. The increase was substantial enough that developers should calculate costs using the actual processing period and cache behavior.
This chart records the August announcement, not a promise of today's rates. See the DeepSeek update log and current pricing before budgeting. The later V4.1 Flash release brings another change to the model lineup.
Adapted from the original Chinese article, published on August 14, 2026.
Comments NOTHING