Tencent Hunyuan officially launched Hy4 Flash, a 224B-parameter model activating 12B per token with multimodal input and 1M-token context. Through coordinated pretraining, post-training and inference optimizations including hybrid attention, LatentMoE, Engram lookup parameters and 4-bit quantization, its per-task cost is just 1/20 of Hy4 and 1/8 of DeepSeek-V4.1-Flash, with official claims of slightly better overall quality than Hy4 preview and plans to open-source.
Modelsslug: hy4-flash
Hy4 Flash
Tencent Hunyuan's cost-efficient model with 224B total and 12B active parameters, multimodal and 1M context.
Curated Intelligence
IT之家 · 智能时代T2·Models
78 pts
腾讯混元 (Tencent Hunyuan)T1·Models
78 pts
Tencent has released Hy4, its new-generation flagship model, now available via Tencent Cloud TokenHub and OpenRouter alongside published benchmark results. The company also previewed Hy4 Flash, a cost-efficiency-focused variant promising faster inference and tighter product co-design.