Background
Tencent announced Hy4 flash on November 16, 2026.[1] The company positions it as its most cost-effective Hy model to date, aimed at everyday workflows.[1]
What happened
Hy4 flash has 224 billion total parameters and 12 billion active parameters, with a context length of up to 1M tokens.[1] Tencent says it cuts per-task cost to 1/20 of Hy4 on its internal WorkIndex benchmark.[1]
Key details
Tencent gives approximate API prices of US$0.070 per million input tokens and US$0.281 per million output tokens.[1] The model is available through Tencent Cloud TokenHub, WorkBuddy, Yuanbao, WeCom, QQ, ima and Marvis.[1]
The model is also positioned against the larger Hy4 line. Tencent says Hy4 flash is the cheaper option in that comparison, while keeping the million-token context window.[1]
Open source
Tencent says it plans to open-source Hy4 flash, with no release date given.[1]
Impact
The announcement gives developers and enterprises a lower-cost Tencent model with a long context window, with the cost figures drawn from Tencent's own internal benchmark.
Sources
- 1
Cite this page: Tencent launches Hy4 flash, a cheaper 1M-context multimodal model. TellWorld, https://www.tellworld.net/news/hy4-flash-2026. Updated 2026-11-18