Tencent Releases Hy4 Flash AI Model
- •Tencent releases Hy4 flash with 224 billion total parameters and 12 billion active parameters
- •Internal WorkIndex results show per-task costs at 1/13 of Hy4 Preview and 1/20 of Hy4
- •Blind test spans 85 WorkBuddy users, 259 tasks and 27 categories; availability includes seven Tencent services
Tencent announced Hy4 flash on November 17 as the latest model in its Hy family, with 224 billion total parameters, including 64 billion Engram parameters, and 12 billion active parameters. It accepts multimodal inputs and supports context up to 1M tokens. Tencent prices it at approximately US$0.070 per million input tokens, US$0.281 per million output tokens and US$0.003 per million cached input tokens, describing it as the most cost-effective Hy model to date and a substantial price reduction from Hy4.
The model uses hybrid attention, LatentMoE and Engram architectural features; its KV cache is 1/24 the size of Hy3’s and 1/6 that of Hy4’s. Tencent says reinforcement learning reduced average end-to-end task output tokens by 50 percent. For evaluation, the team created WorkIndex, an internal benchmark based on typical WorkBuddy tasks, and paired it with human assessments of deliverable quality, single-turn output length and task turnaround time.
Tencent says optimization using megakernel, 4-bit quantization and DFly techniques enables decoding speeds above 150 tokens per second. On WorkIndex tasks involving PowerPoint creation and editing, document editing, spreadsheet repair, audio/video processing and visual understanding, Hy4 flash slightly outperformed Hy4 Preview, which has 770 billion total parameters and 49 billion active parameters. Its per-task cost was 1/13 of Hy4 Preview’s, 1/20 of Hy4’s and 1/8 of DeepSeek-V4.1-Flash-1017’s Max tier at peak-hour pricing.
In a blind evaluation, 85 internal Tencent WorkBuddy users tested 259 tasks across 27 categories. Tencent reports satisfaction close to Claude Haiku 5.5, the lowest per-task cost among the evaluated models and the shortest completion time. Hy4 flash is available through Tencent Cloud TokenHub, WorkBuddy, Yuanbao, WeCom, QQ, ima and Marvis, with OpenRouter integration planned. Through November 30, users can access it free on WorkBuddy and receive 50 percent off standard pricing on Tencent Cloud TokenHub. Tencent plans to open-source the model and recommends it for everyday productivity tasks; it points users to Hy4 for complex work such as long-horizon software engineering.
The fastest, most reliable AI news, straight to your inbox