Yes — **DeepSeek V4 Pro is live on the DeepSeek API today** (as of 2026-08-13), as the production/GA release. **What I verified on DeepSeek's official API docs:** - The model ID **`deepseek-v4-pro`** is listed as a current, callable model — "simply use `deepseek-v4-flash` or `deepseek-v4-pro` to access the latest version" ([DeepSeek API Docs – Quick Start](https://api-docs.deepseek.com/)). - The current build is designated **DeepSeek-V4-Pro-0813** — the footnote states "the deepseek-v4-pro model has been updated to DeepSeek-V4-Pro-0813." - Live pricing on the official [Models & Pricing page](https://api-docs.deepseek.com/quick_start/pricing): **$0.435 / 1M input tokens (cache miss)**, **$0.87 / 1M output tokens** ($0.003625 cache hit), **1M-token context**, **384K max output**, supports thinking mode, JSON output, tool calls, Responses API and Anthropic API format. - The official docs' example code calls `model="deepseek-v4-pro"` against `https://api.deepseek.com/chat/completions`, confirming it's usable right now. **Corroboration on the GA release date:** - OpenRouter's listing for `deepseek/deepseek-v4-pro-0813` describes it as "**the GA release of DeepSeek V4 Pro**," **released Aug 12, 2026**, at the same $0.435/$0.87 pricing and 1M context ([OpenRouter – DeepSeek V4 Pro 0813](https://openrouter.ai/deepseek/deepseek-v4-pro-0813)). One small note: the deepseek.com homepage still shows an older banner about the V4-Flash public beta ("V4-Pro version remains unchanged for now"), but that banner predates the 0813 update — the authoritative API docs already list the V4-Pro-0813 build as the current version. So you can call `deepseek-v4-pro` via the API today.