DeepSeek moves V4-Pro into general availability
The August release added reasoning controls and native Responses support, alongside a new peak and off-peak pricing structure.
THE STORY IN THREE POINTS
- V4-Pro reached app, web and API in August.
- The release added low, high and max reasoning effort.
- The announced off-peak rates were half of peak rates.
The announcement
DeepSeek announced V4-Pro general availability on August 13 for its app, website and API. The release added low, high and max reasoning effort controls for V4-Pro and V4-Flash, together with native support for the Responses API format. DeepSeek also announced peak and off-peak API pricing, effective August 16 at 16:00 UTC. It attributed stronger agent results to the updated model.
Editorial context
Reasoning controls create a scheduling and quality choice for developers. A short extraction job and a difficult software repair need not use the same settings. The relevant comparison is the total effort to obtain a correct result, including retries and review, because a faster individual response may still leave more work for a person.
Editorial context: what to watch
A practical evaluation could run representative tasks at each effort level and record success, latency and token use. Flexible batch work could be assessed separately from interactive requests. The release announcement establishes the historical change; current model availability and billing should be checked against today’s documentation before committing a production budget or a migration schedule.
Sources: DeepSeek: V4-Pro GA release, August 13 ↗
