DeepSeek-V4-Pro Officially Released: 1M Context, API Pricing Increased

APPSO · wechat · 2026-08-12

The official version of DeepSeek's model, DeepSeek-V4-Pro-0813, is now live. The new version supports a 1M context length and 384K max output, with thinking mode enabled by default. It is fully compatible with both OpenAI and Anthropic API formats, further extending its reach into Agent and Coding scenarios.

The previously announced API price hike has also taken effect: V4-Pro's cache-miss input and output prices are set at 3 RMB and 6 RMB per million tokens respectively, which is 3 times the price of the V4-Flash version. The product line structure is now clearer, with Flash targeting low-cost, high-concurrency scenarios (limit 2500), while Pro aims at high-spec complex tasks (concurrency limit 500).

Related event: DeepSeek V4 Pro Officially Released with Disruptive Pricing and 1M Context(12 posts)→

Original post →

More from Models

Models channel →