Tencent Cloud (腾讯云) Raises Hunyuan Model Token Prices, Some Fees Jump More Than 5x

Price change and scope

Tencent Cloud (腾讯云) has announced new paid tiers and pricing adjustments for several models on its intelligent agent development platform, including the commercial rollout of previously beta models. The company said GLM 5, MiniMax 2.5 and Kimi 2.5 are exiting public beta and moving to paid plans. Most notably, the Tencent HY2.0 Instruct (Hunyuan 2.0) model’s input price has been raised from ¥0.0008 to ¥0.004505 per 1,000 tokens, and its output price from ¥0.002 to ¥0.01113 per 1,000 tokens — increases of more than fivefold in both directions.

Why now and product context

Why the jump? Tencent appears to be shifting from a free or subsidized beta posture to full commercial pricing as it scales AI products and absorbs inference costs. The move coincides with a push to productize agent-style services: Tencent recently launched WorkBuddy, an all-scenario AI agent reportedly compatible with OpenClaw-style “Xiaolongxia” skills, and it has been reported that the Tencent PC security team is testing a similar agent called QClaw with claimed WeChat integration. It has also been reported that Tencent is developing a “top-secret” WeChat AI agent slated for limited testing mid-year.

Geopolitics, competition and implications

This pricing change comes as Chinese cloud vendors race to monetize large models amid intensifying global scrutiny of AI technology, export controls and chip supply constraints that make domestic scale and revenue more important. For Chinese developers and enterprises, higher per-token costs will change economics for high-volume deployments. For competitors and international observers, the move signals Tencent’s pivot from experimentation toward broad commercialization of its Hunyuan family — and raises the question: will users absorb the cost, switch vendors, or optimize models to avoid sticker shock?