How much is the DeepSeek V4 Pro API price increasing?

The increase depends on token type and time of day. V4 Pro cache-hit input rises by 6x off-peak and 12x at peak, cache-miss input by 1.5x and 3x, and output by 2.25x and 4.5x.

When do the new DeepSeek V4 API prices take effect?

The peak and off-peak rates take effect at 00:00 Beijing time on August 17, 2026. Peak hours are 09:00-12:00 and 14:00-18:00 Beijing time; all other hours are billed at half the peak rate.

Do existing DeepSeek V4 Pro API integrations need a new model name?

No. DeepSeek keeps the API model name deepseek-v4-pro. Its official pricing page now identifies the backend version as DeepSeek-V4-Pro-0813.

DeepSeek Raises API Prices by Up to 12x With the Official V4 Pro Release

DeepSeek V4 Pro 0813 is now official. On August 17, DeepSeek will introduce peak and off-peak API pricing, with V4 Pro rates rising by as much as 12 times the previous price.

DeepSeek's largest API price increase lands on cache-hit input for V4 Pro. During peak hours, the price will rise from RMB 0.025 to RMB 0.30 per million tokens - a 12-fold increase. Even the off-peak rate will be six times the previous price.

The new schedule takes effect at 00:00 Beijing time on August 17, 2026, four days after DeepSeek listed the official DeepSeek-V4-Pro-0813 build behind the existing deepseek-v4-pro API name.

The new peak and off-peak schedule

Peak hours are 09:00-12:00 and 14:00-18:00 Beijing time. Every other hour is off-peak, and DeepSeek prices those periods at half the peak rate.

For V4 Pro, off-peak usage will cost RMB 0.15 per million cache-hit input tokens, RMB 4.50 per million cache-miss input tokens, and RMB 13.50 per million output tokens. Peak rates double to RMB 0.30, RMB 9.00, and RMB 27.00.

V4 Flash follows the same schedule at lower absolute prices. Its off-peak rates will be RMB 0.05 for cache-hit input, RMB 1.50 for cache-miss input, and RMB 4.50 for output. Peak rates will be RMB 0.10, RMB 3.00, and RMB 9.00.

The increases are uneven

Cache-hit input carries the sharpest increase. V4 Pro rises by 6x off-peak and 12x at peak. V4 Flash rises by 2.5x off-peak and 5x at peak.

Cache-miss input rises more moderately: 1.5x off-peak and 3x at peak for both models. Output pricing rises by 2.25x off-peak and 4.5x at peak.

That makes cache behavior and request timing much more important to the final bill. Workloads that can run outside the two peak windows will pay half the peak price, but every new off-peak rate is still higher than the current rate.

What developers should check before August 17

Teams using DeepSeek in production should separate cache-hit input, cache-miss input, and output tokens in their cost model. A single blended token estimate will hide the difference between a 1.5x increase and a 12x increase.

Flexible batch jobs should be moved to off-peak hours where possible. Long agent sessions should also be tested for cache reuse, because repeated cache misses will now cost RMB 4.50 off-peak or RMB 9.00 at peak for every million V4 Pro input tokens.

V4 Pro 0813 is now the official API build

DeepSeek's official model table now maps deepseek-v4-pro to DeepSeek-V4-Pro-0813. Existing integrations keep the same model name and base URL. The release also retains the 1-million-token context window, 384K maximum output, thinking and non-thinking modes, tool calls, and native Responses and Anthropic API support.

DeepSeek's benchmark table shows broad agent and coding gains over the Pro Preview build. The figures are vendor-reported and depend on each test's harness and settings, but they establish that 0813 is a substantial post-training update rather than a simple version rename.

The immediate issue for API users is the price reset. The 0813 backend is already listed, and the new billing schedule begins on August 17.

DeepSeek-V4 will launch very soon, either this afternoon or tonight.