What Tencent announced
On 16 November 2026, Tencent published Introducing Hy4 flash: Ultimate cost-efficiency for everyday productivity on the official Hy research site. The page presents Hy4 flash as Tencent's most cost-effective Hy model to date. It states that the model has 224 billion total parameters, including 64 billion Engram parameters, with 12 billion active parameters per token. Like Hy4, it is natively multimodal and has a 1-million-token context window.
Official API prices on that page, per million tokens, are $0.003 for cached input, $0.070 for input, and $0.281 for output. The same table lists Hy4 at $0.042 cached, $0.834 input, and $2.501 output. Tencent says reinforcement learning cut average end-to-end output tokens by 50%, and that combining megakernels, 4-bit quantization, and DFly pushes decoding speed past 150 tokens per second.
How Tencent positions it, and where it is available
The research page says Hy4 flash is slightly above Hy4 preview in overall task performance. On WorkIndex, an internal benchmark of typical WorkBuddy tasks, Tencent says the average cost per task is 1/13 of Hy4 preview, 1/20 of Hy4, and 1/8 of DeepSeek-V4.1-Flash-1017 (Max tier, peak-hour pricing). In a blind evaluation with 85 internal WorkBuddy users across 259 tasks in 27 categories, Tencent says user satisfaction was close to Claude Haiku 5.5, with the shortest end-to-end completion time and a lower cost per task than any other evaluated model in that test. Those results are Tencent's own evaluation, not an independent audit.
The official @TencentHunyuan account, in a post published at 06:38 UTC on 16 November 2026, described Hy4 flash as matching Claude Haiku 5.5 (high) on daily work in human blind tests, with an 18x task cost reduction. The research page does not state an 18x figure; it uses the wording and WorkIndex ratios above.
Tencent says Hy4 flash is now available on Tencent Cloud TokenHub, WorkBuddy, Yuanbao, WeCom, QQ, ima, Marvis, and more. Through 30 November 2026, the page says users can access it for free in WorkBuddy and get 50% off API pricing via Tencent Cloud TokenHub. The research page, as rendered on 17 November 2026, still said the model will launch on OpenRouter soon. Later the same day, at 16:01 UTC, the official @TencentHunyuan account said Hy4 flash was live on OpenRouter with a limited-time 50% off rate of $0.035 input, $0.1405 output, and $0.0015 cache read per 1 million tokens. This article does not use OpenRouter.
What Tencent recommends, and what is still open
Tencent recommends Hy4 flash for most everyday productivity tasks, such as deep search, document processing, slide creation, simple frontend demos, and computer use. It continues to recommend Hy4 for complex workloads such as long-horizon software engineering, which it says has a higher intelligence ceiling. The company also says it plans to release Hy4 flash as an open-weight model. No official Hugging Face repository for those weights was confirmed at the time of writing.
Go to the source
Sources & further reading
Based on the official documentation below. Features and access can change; check the provider before making a decision.
- Tencent Hy official research page: Introducing Hy4 flash (dated 16 November 2026), rendered 17 November 2026 (opens in a new tab)
- Official @TencentHunyuan post announcing Hy4 flash, 16 November 2026, 06:38 UTC (opens in a new tab)
- Official @TencentHunyuan post on OpenRouter availability and 50% off rates, 16 November 2026, 16:01 UTC (opens in a new tab)
Was this page useful?
Help us make our coverage more useful. This is feedback on the page, not a rating of the tool.
Go a little further