DeepSeek V4 Flash + Codex: Too Fast for Cheap
Try DeepSeek V4 Flash via Codex for ultra-fast Agent performance at just ¥1 per million tokens—budget price, Pro-level work.
“AI Disruption” Publication 10,400 Subscriptions 20% Discount Offer Link.
Conclusion first: On July 31, DeepSeek quietly released the official API for V4-Flash. As soon as the news came out, I connected it via Codex and gave it a try. The response speed is really fast, and the input price is only ¥1 per million tokens—a budget version whose Agent working ability actually surpassed their own more expensive Pro preview version. Ordinary people can now play with Agents at low cost.
💡 A cold splash of water first: This is only an API interface update. You still can’t use this “official version” in the DeepSeek App or on the web today—the experience is the same as yesterday. If you just want to chat, wait for the official V4-Pro version on the App in early August. But if you’re willing to connect it via Codex (even ordinary people like me can handle it), you can use it right now.
