Skip to content

An editorial team and a community building with AI

vibecoding.incJoin Discord ↗
← All articles

News · DeepSeek · 2 min read

Event date: · Published:

DeepSeek makes DeepSeek-V4-Pro generally available and adds off-peak pricing

On August 13, 2026, DeepSeek made DeepSeek-V4-Pro generally available on its app, web and API and published MIT-licensed weights. Peak and off-peak pricing took effect on August 16, with off-peak at half the peak price.

Archive entry: this covers an event from August 13, 2026, and we published it on November 8, 2026. The details come from sources dated at the time and may have changed since.

On August 13, 2026, DeepSeek made DeepSeek-V4-Pro generally available on its app, web and API, as described in the DeepSeek API release post. The API served the model as deepseek-v4-pro, and the app and web called it Expert Mode.

What changed

  • Weights: DeepSeek published DeepSeek-V4-Pro-0813 on Hugging Face under the MIT license, with about 1.65T parameters.
  • API: The API natively supported the OpenAI Responses API format and was adapted for Codex with a one-click script. Thinking effort for V4-Pro and V4-Flash could be set to low, high or max.
  • Pricing: peak and off-peak pricing took effect at 16:00 UTC on August 16, 2026, with off-peak at half the peak price.
  • Rates: the pricing page listed, per million tokens for DeepSeek-V4-Pro-0813 (off-peak / peak), cache hit $0.022 / $0.044, cache miss $0.66 / $1.32 and output $1.98 / $3.96.
  • Benchmarks: listed coding scores included Terminal Bench 2.1 at 87.9, DeepSWE at 62.7 and Toolathlon-Verified at 74.1.

What it meant for people building with AI

Developers could run DeepSeek's top model from Responses API clients such as Codex, and cut costs by calling it off-peak, when the price was half the peak rate. The pricing page also listed concurrency of 500 and no vision support, two limits to weigh against a given workload.

Where the details were published

The weights were on the Hugging Face model page, while the DeepSeek API change log and the pricing page recorded the update and the rates. On August 14, 2026, Alibaba Cloud Model Studio listed the model with 1.6T total and 49B active parameters and up to 1M context.

Sources

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.

DeepSeek makes DeepSeek-V4-Pro generally available and adds off-peak pricing | vibecoding.inc