Tencent has launched Hy4 Flash, a lower-cost version of its flagship Hy4 model aimed at everyday work such as research, slide creation and spreadsheet editing. The model went live on Monday, November 16, less than two weeks after the flagship Hy4 arrived on November 4.
Hy4 Flash is built for volume rather than peak capability. It has 224 billion total parameters, with 12 billion activated per inference, compared with 770 billion total parameters and 49 billion activated in the full Hy4 model. It supports multimodal inputs and a context window of up to 1 million tokens, matching the flagship on context length.

What Was Announced?
The headline claim is cost. On Tencent’s internal WorkIndex benchmark, Hy4 Flash cost one-twentieth as much per task as the full Hy4 model, a 95 percent reduction. The South China Morning Post reported that the flash model also costs about one-eighth as much per task as DeepSeek’s V4.1-Flash-1017. These are vendor-reported figures from Tencent’s own testing, not independent benchmarks, and no third-party leaderboard scores for Hy4 Flash have been published yet.
Tencent also ran a blind test in which 85 WorkBuddy users rated outputs across 259 tasks. The company said Hy4 Flash scored close to Claude Haiku 5.5 on user satisfaction while posting the lowest cost and shortest completion time among the models tested.
Pricing and Availability
API list prices are $0.07 per million input tokens, $0.281 per million output tokens and $0.003 per million cached input tokens, according to TechNode. In yuan terms, ChainCatcher reported prices of 0.5 yuan per million input tokens and 2 yuan per million output tokens, roughly 12 times and 9 times cheaper than Hy4’s listed API rates.
The model is available now in Tencent’s WorkBuddy office agent and its Yuanbao consumer assistant, and through Tencent Cloud’s TokenHub. WorkBuddy users can access it free through November 30, and TokenHub is offering API access at half the list price. Tencent plans to release the model’s weights and to make it available on OpenRouter, though no date has been given for either step.
Why It Matters
Hy4 Flash completes the two-tier Hy4 lineup Tencent previewed on October 30 at its Global Digital Ecosystem Summit: a flagship model for demanding agentic work and a cheaper variant for routine tasks. The flagship Hy4 launched on November 4 with native vision understanding and computer-use capabilities, and Tencent has promised open weights for it as well.
The launch lands on the same day Alibaba released its flagship Qwen4-Max, underlining how quickly China’s leading labs are now shipping cheaper, faster variants alongside their flagships. For developers, the practical question is whether Hy4 Flash’s quality holds up outside Tencent’s own tests. The blind test against Claude Haiku 5.5 is encouraging, but independent evaluations will matter more than internal benchmarks, and the promised open weights will let the wider community judge the model for itself.

Sources
- TechNode: Tencent launches Hy4 flash, cutting AI task costs by 95% in internal tests
- South China Morning Post: Alibaba says new flagship AI model cuts task costs by 60% while tightening cyber guardrails
- KuCoin / ChainCatcher: Tencent Hunyuan Launches Hy4 Flash with 224B Total Parameters and 12B Activated Parameters

