DeepSeek V4 Pro 0813 Is Live as DeepSeek Warns of a Major API Price Increase
DeepSeek V4 Pro 0813 went live at about 11 p.m. Beijing time on August 13. The model keeps the same API name and current rates, but DeepSeek is warning users to expect a significant price increase soon.
DeepSeek V4 Pro 0813 went live at about 11 p.m. Beijing time on August 13, 2026. The release arrived quietly. DeepSeek changed the model version on its official pricing page to DeepSeek-V4-Pro-0813, while the API model name stayed deepseek-v4-pro.
That is the useful part for developers already using the Pro API. Existing integrations pick up the 0813 backend without changing the model parameter or base URL.
The price warning is the other half of the release
DeepSeek's current rates have not changed. V4 Pro still costs $0.003625 per million cached input tokens, $0.435 per million uncached input tokens, and $0.87 per million output tokens.
A new footnote on the same page matters more than those numbers. DeepSeek says it plans to raise API prices in the near future and expects the increase to be significant. It has not published the new rates or an effective date. Anyone budgeting a production workload should treat today's price as current, not permanent.
The warning follows reports of slower API performance after the July 31 V4 Flash release brought a surge of usage. DeepSeek has not publicly said that a compute shortage is the reason for the planned increase, so that explanation should remain a reported interpretation rather than an official fact.
The API contract is familiar
The official model page lists a 1-million-token context window and a maximum output length of 384K tokens. V4 Pro supports thinking and non-thinking modes, JSON output, tool calls, the Responses API, the Anthropic API, chat prefix completion, and FIM completion in non-thinking mode.
DeepSeek has also added a growing list of coding-agent guides around those two API formats. Codex can use the Responses API route, while Claude Code and other Anthropic-style tools can connect through DeepSeek's Anthropic endpoint. The 0813 rollout is a backend update, not an integration project.
Grok 4.6 landed in the same launch window
The timing gives developers an unusually direct choice. SpaceXAI released Grok 4.6 on August 12, one day before the DeepSeek update in Beijing. Grok 4.6 is aimed at long-running agents, coding, and more ambitious visual or interactive work. Its API starts at $2 per million input tokens and $6 per million output tokens, far above DeepSeek's current V4 Pro rates.
SpaceXAI reports that Grok 4.6 improves on Grok 4.5 across its published agent and knowledge-work tests. DeepSeek and SpaceXAI did not use the same benchmark package or harness, however. The two release tables cannot settle which model is better. They do make the competitive question more interesting: how much quality and reliability does each API deliver for its price once DeepSeek's new rates arrive?
What to watch next
The next official DeepSeek notice should answer two practical questions: the new token rates and when they take effect. Until then, the sensible move is to measure V4 Pro 0813 on real workloads, keep cache-hit rates visible, and avoid building a long-term cost forecast around the current price.
The release itself is straightforward. V4 Pro 0813 is live, existing API calls continue to work, and today's rates still apply. The uncertainty sits in the bill that comes next.
Sources
Frequently asked
Is DeepSeek V4 Pro 0813 live?
Yes. DeepSeek's official models and pricing page lists the API model version as DeepSeek-V4-Pro-0813. Existing calls continue to use the deepseek-v4-pro model name.
Has DeepSeek increased V4 Pro prices?
Not yet. The current V4 Pro rates remain $0.003625 per million cached input tokens, $0.435 per million uncached input tokens, and $0.87 per million output tokens. DeepSeek says a significant increase is planned but has not published the new rates or date.
Is DeepSeek V4 Pro 0813 cheaper than Grok 4.6?
At the prices published on August 13, yes. Grok 4.6 starts at $2 per million input tokens and $6 per million output tokens, while DeepSeek V4 Pro is currently $0.435 for uncached input and $0.87 for output. DeepSeek has warned that its API pricing will rise.