Tencent Cloud Slashes DeepSeek-V4 Model Prices by Up to 97.5%
Miles Bennett
Tencent Cloud will slash API pricing for the DeepSeek-V4 model family on June 3, with the steepest cut reaching 97.5% — a signal that large-model APIs are shifting from premium tools to commodity infrastructure.
What is being cut, and by how much?
Tencent Cloud's AI-agent development platform will apply new pricing from June 3, 2026, 00:00 Beijing time.
The cuts target the DeepSeek-V4 model family, with reductions of up to 97.5%.
In plain terms = a call that cost 100 units now costs as little as 2.5 — two orders of magnitude gone.
Does the service shrink with the price?
Tencent Cloud stated explicitly: only pricing changes; model capabilities remain unchanged.
This means → users get the same model and the same inference quality — just a thinner bill.
For developers, the cost ceiling drops sharply, opening trial-and-error room for smaller teams.
What signal does this price war send?
A 97.5% cut goes far beyond routine adjustment — it reads as strategic pricing to lock in developer ecosystems.
This reflects a full-blown price war among Chinese cloud providers around large-model APIs, with compute being subsidised as a traffic on-ramp.
In plain terms = the playbook is no longer "sell compute for margin" but "acquire developers first, monetise value-added services later."
Content is for reference only, not financial advice.