Tencent launches Hy4 Flash, a lower-cost multimodal model with a one-million-token context
Reported by techflowpost, chaincatcher, tokenpost.com and 10 more outlets13 sources ↓
Tencent launched Hy4 Flash on November 16, positioning it as a cheaper option for everyday work. The model supports multimodal inputs and a one-million-token context; Tencent’s listed API prices are lower than Hy4’s.
Key points
- The model has 224 billion total parameters, with 12 billion active per inference
- Tencent lists cached input at RMB 0.02 per million tokens
- It is available through Tencent Cloud TokenHub, WorkBuddy and Yuanbao
- Tencent is offering a 50% TokenHub API discount through November 30, 2026
- Tencent plans to release the model weights and add OpenRouter availability
Sources
- techflowpostRead ↗Tencent Hunyuan Launches Hy4 Flash, Supports 1M Context
- chaincatcherRead ↗Tencent Hunyuan released Hy4 flash, with total parameters of 224B and activation of 12B
- tokenpost.comRead ↗Tencent’s Hy4 Flash Cuts Automated-Task Cost to 0.038 Yuan
- kucoinRead ↗Tencent Hunyuan Launches Hy4 Flash with 224B Total Parameters and 12B Activated Parameters
- TechNodeRead ↗Tencent launches Hy4 flash, cutting AI task costs by 95% in internal tests
Details
Tencent says Hy4 Flash is designed for tasks such as research, document editing, slide creation and spreadsheet work. The model accepts multimodal inputs and supports a context window of up to one million tokens. It has 224 billion parameters in total, including 64 billion Engram parameters, and activates 12 billion per inference.
Tencent’s listed API prices are RMB 0.5 per million input tokens, RMB 2 per million output tokens and RMB 0.02 per million cached input tokens. Unite.AI reports these rates are about 92% lower for input and 89% lower for output than Hy4’s list prices. Tencent Cloud TokenHub is offering a 50% API discount through November 30, 2026; WorkBuddy access is also free through that date, according to Unite.AI and TechNode.
Tencent’s internal evaluations found lower per-task costs than its other models. TechNode reports that Hy4 Flash cost one-twentieth as much per task as Hy4 on Tencent’s WorkIndex benchmark; Unite.AI reports the same comparison and says Tencent measured one-eighth the cost of DeepSeek V4.1 Flash at peak-hour pricing. In a separate test of 259 tasks by 85 WorkBuddy users, Tencent reported satisfaction close to Claude Haiku 5.5; Crypto Briefing gives the scores as 2.79 for Hy4 Flash and 2.81 for Claude Haiku 5.5.
The model is available through TokenHub, WorkBuddy and Yuanbao, with availability through OpenRouter planned. Tencent also plans to release its weights.
Written from coverage as of
Images
- techflowpost
- TechNode
- Unite.AI
- orcarouter.ai
Stocks in this story
Change since open: 20 Nov to 21 Nov 2026
Last refresh