Tencent launches Hy4 flash, a low-cost multimodal model with a 1M-token context
Tencent released Hy4 flash on November 17, 2026, a 224-billion-parameter model with 12 billion active parameters, multimodal input and a 1M-token context, priced at about $0.07 per million input tokens and $0.281 per million output tokens, its cheapest Hy model so far.
What happened
- Hy4 flash has 224 billion total parameters (including 64 billion Engram parameters), activates 12 billion per token, accepts multimodal input and supports up to 1M tokens of context.
- Tencent lists pricing at about $0.070 per million input tokens, $0.281 per million output tokens and $0.003 per million cached input tokens.
- On its internal WorkIndex benchmark of everyday WorkBuddy tasks, Tencent says Hy4 flash slightly outperforms the larger Hy4 Preview at about 1/13 of the per-task cost.
- It is available on Tencent Cloud TokenHub, WorkBuddy, Yuanbao, WeCom, QQ, ima and Marvis, with OpenRouter support listed as coming soon; it is free on WorkBuddy and 50% off on TokenHub through November 30.
- Tencent recommends it for everyday work such as research, documents, slide decks and computer use, keeps the larger Hy4 for long software projects, and says it plans to open-source Hy4 flash.
Why it matters: Hy4 flash pushes the price of a long-context, multimodal model for routine office and agent tasks very low. Teams paying for high-volume document or research work can compare it once it reaches OpenRouter, keeping in mind that the quality claims come from Tencent's own tests.
Sources
- Tencent via PR Newswire: Tencent Launches Hy4 flash prnewswire.com
Checked November 17, 2026 against the sources above. Claims about performance or benchmarks are the company's own unless stated otherwise.