DeepSeek V4 Pro Launch: Pricing, API Compatibility, and Performance Insights
The article announces the quiet release of DeepSeek V4 Pro (version 0813), details its token pricing and cache‑hit cost advantages over V4‑Flash, highlights its near‑Fable 5 performance, describes its dual OpenAI‑compatible and Anthropic APIs, and shares resources for AI learning and project integration.
The author, who follows large‑model developments for AI projects, notes that different models vary in accuracy and price.
DeepSeek V4 Pro (DeepSeek‑V4‑Pro‑0813) was quietly released; the official announcement is absent, but the version number on the pricing page has been updated and the API is already callable, turning the preview into a full release.
Pricing per 1 M tokens is disclosed: cache‑hit input costs 0.025 yuan, cache‑miss input 3 yuan, and output 6 yuan. By contrast, V4‑Flash charges 0.02 yuan, 1 yuan, and 2 yuan respectively, meaning the higher cost of V4 Pro is confined to new content, while repeated prompts with cache hits are extremely cheap.
In several benchmarks DeepSeek‑V4‑Pro‑0813 approaches the performance level of Fable 5, representing a noticeable improvement over the preview version.
The model is offered through two API styles: an OpenAI‑compatible Responses API and an Anthropic API . This allows projects that already use OpenAI‑style calls or Anthropic SDKs to switch by merely changing the base URL, without rewriting code or adding adapters.
The author also mentions that local deployment is costly; a GPU‑equipped server was purchased for over 200 k CNY.
Additional resources are provided, including links to an AI + full‑stack learning roadmap, five AI + RAG + Agent projects, and a detailed guide for integrating Codex with DeepSeek‑V4‑Flash.
Overall, the piece combines a product release analysis with practical guidance for developers interested in leveraging the new model.
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
SpringMeng
Focused on software development, sharing source code and tutorials for various systems.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.