Tencent's Hunyuan Hy4 flash model exceeded 50 trillion tokens in its first day of use
Read more
TechNode
technode.com

Tencent's Hunyuan Hy4 flash model exceeded 50 trillion tokens in its first day of use

Tencent announced that its Hunyuan Hy4 flash model processed over 50 trillion tokens in just the first day after its launch on November 16. This figure was the highest among all models in the Hunyuan series for their first day of operation.

The strong start of Hy4 flash also reflected in its performance across various platforms. On November 17, the model ranked first in the daily LLM rating on OpenRouter.

Furthermore, Tencent noted that on its WorkBuddy platform, Hy4 flash demonstrated ten times greater usage compared to its predecessor, Hy4 preview, and helped complete over 50 million real user tasks.

According to Tencent data, Hy4 flash utilizes a hybrid activation architecture and has a total of 224 billion parameters, including 64 billion Engram parameters, with 12 billion parameters activated during inference. The model also supports multimodal processing and a context window of up to 1 million tokens, achieving a generation speed of about 200 tokens per second.

Of particular interest is the balance between performance and cost offered by Hy4 flash. According to Tencent's pricing, the model costs only 0.5 yuan ($0.07) per million input tokens and 2 yuan ($0.3) per million output tokens, while cached input tokens cost only 0.02 yuan ($0.003) per million.

Based on Tencent's internal WorkIndex benchmark, which assesses standard office tasks, Hy4 flash costs only one-thirteenth compared to Hy4 preview and one-twentieth compared to Hy4 for completing a single task. Its cost is also approximately one-eighth of the cost of mainstream lightweight models in the same category.

These advantages make the model particularly suitable for frequently performed office operations such as creating and editing presentations, document processing, spreadsheet recovery, audio and video processing, and visual understanding.

The launch followed a previous surge in demand for Hy4. When the model first appeared on WorkBuddy on November 4, demand driven by its agent capabilities exceeded expectations, and many users reached usage limits on the very first day. In response, Tencent increased Hy4 inference clusters tenfold and expanded free usage quotas.

However, this time the team apparently prepared more thoroughly for high competition before the launch of Hy4 flash, and there are currently no reports of widespread usage limit issues.

This development also reflects a broader shift in the competitive landscape of the artificial intelligence industry. As daily token usage in China continues to grow, tokens are becoming an increasingly important metric for measuring AI adoption. Meanwhile, local authorities across the country are introducing new measures, ranging from computing power and model vouchers to token-based funding, to make AI services more accessible.

Competition is now determined not only by the number of parameters or model performance. Achieving strong results while maintaining low operating costs is becoming equally important. The debut of Hy4 flash with 50 trillion tokens sends a clear signal: the market needs not only more powerful AI models but also models that are accessible enough for frequent daily use.

Nevertheless, there remains a gap between high initial usage and long-term commercial value. Competition among lightweight AI models may increasingly boil down not only to prices and token usage. The real challenge is to provide solutions for real business needs stably and reliably while keeping overall costs low. For Tencent's Hunyuan, the strong debut of Hy4 flash is a promising start, but its ability to convert initial usage momentum into sustainable user value will be a key test in the future.

Similar stories

Tencent releases Hy4 flash model weights for local deployment
Read more
technode.com

Tencent releases Hy4 flash model weights for local deployment

Tencent announced the weights of its artificial intelligence model, Hy4 flash, on Wednesday. This release came two days after the model was introduced to users via the company's applications and cloud API. Developers can now download the standard version of the model, its FP8 format version, as well as the base model for further training.

This release aligns with Tencent's previous statement regarding plans to make the model weights publicly available. The Hy4 flash model is capable of processing both text and image input, and its context window reaches one million tokens.

The model has approximately 224 billion parameters, with about 12 billion parameters activated per token. The main goal of Hy4 flash is to improve daily performance in tasks such as document preparation, research, and spreadsheet work. Tencent accompanied the weights with relevant deployment materials, distributing them under the Tencent Hy license.

Tencent releases Hy4 flash model with 224 billion parameters for 2 yuan per million output tokens
Read more
pandaily.com

Tencent releases Hy4 flash model with 224 billion parameters for 2 yuan per million output tokens

The Hunyuan team at Tencent introduced the Hy4 flash model on November 16th, calling it the most economical model in the Hy series to date. This model has a total of 224 billion parameters, including 64 billion Engram parameters, activates 12 billion per token, supports multimodal input, and features a context window of one million tokens.

API tariffs are set at 0.5 yuan per million input tokens, 2 yuan per output token, and 0.02 yuan for cache hits, contrasting with the prices for Hy4, which are 6, 18, and 0.3 yuan respectively. Until November 30th, Hy4 flash will be available for free in WorkBuddy and at half price on Tencent Cloud's API TokenHub.

Tencent explains the achieved efficiency by detailing its operation at each stage. Key results are achieved through WorkIndex—an internal benchmark created based on typical WorkBuddy tasks such as creating and editing slides, processing documents, correcting spreadsheets, handling audio and video, and visual understanding, supplemented by human evaluations. WorkIndex is not publicly available, and Tencent emphasizes that it focuses on daily work rather than public ranking results.

According to Tencent data, in a blind test conducted by 85 internal WorkBuddy users across 259 tasks in 27 categories, the satisfaction level with the Hy4 flash model approached that of Claude Haiku 5.5, demonstrating the lowest cost per task and the shortest task completion time. However, the Tencent WorkBuddy team noted a slight decrease in the model's performance compared to Hy4 when executing long coding tasks.

The Hy4 flash model is already available in Tencent Cloud TokenHub, WorkBuddy, Yuanbao, WeCom, QQ, ima, and Marvis, with plans for integration into OpenRouter. Tencent recommends using Hy4 flash for deep search, document processing, slide creation, simple frontend pages, and computer tasks, while the Hy4 model is intended for more extensive software development tasks. The company intends to make Hy4 flash open source.

Tencent introduces Hy4 flash model, reducing AI task costs by 95% in internal tests
Read more
technode.com

Tencent introduces Hy4 flash model, reducing AI task costs by 95% in internal tests

Tencent released Hy4 flash on Monday—a more cost-effective artificial intelligence model designed for everyday tasks such as creating presentations, conducting research, and editing spreadsheets.

According to Tencent's internal benchmark, WorkIndex, the cost of executing a single task with the Hy4 flash model was only one twentieth compared to the full Hy4 model. The API prices are set as follows: $0.07 per million input tokens, $0.281 per million output tokens, and $0.003 per million cached input tokens.

The Hy4 flash model supports multimodal input and has a context window of up to one million tokens. It possesses 224 billion parameters and activates 12 billion parameters during direct pass.

In a separate blind test, where 85 WorkBuddy users evaluated results across 259 tasks, Tencent reported that the model achieved a user satisfaction level close to Claude Haiku 5.5, while having the lowest cost and shortest completion time among the tested models.

The model is available in WorkBuddy and Yuanbao, as well as via TokenHub on Tencent Cloud. WorkBuddy users can use it for free until November 30th, and TokenHub offers API access at half the standard price. Tencent plans to publish the model weights and make it available on OpenRouter.

Popular