Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

Tencent Releases Hy4 Flash, a 224B Model Priced at 2 Yuan per Million Output Tokens

Tencent launched Hy4 flash on November 16, a 224-billion-parameter model priced at 2 yuan per million output tokens and pitched at everyday office work. The company says a single task costs roughly one-twentieth of the full Hy4 model, with access through WorkBuddy, Yuanbao and Tencent Cloud's TokenHub.

Published
腾讯发布混元 Hy4 flash:224B 参数模型,输出每百万 token 定价 2 元
Image source: tencent.com

Tencent released Hy4 flash on Monday, a low-cost language model aimed at everyday office work such as research, slide creation and spreadsheet editing. The headline specification is 224 billion total parameters, which keeps the model in large-model territory even as Tencent markets it as a cheap, workhorse tier rather than a capability flagship.

The price is the headline: 2 yuan per million output tokens, and the company says it puts a 224-billion-parameter model into the range where high-volume, multi-step workloads start to make economic sense. On Tencent's internal WorkIndex benchmark, the company says Hy4 flash completes a single task at roughly one-twentieth the cost of the full Hy4 model.

Under the hood, Hy4 flash uses sparse activation, engaging about 12 billion parameters per forward pass so that inference does not carry the full 224B cost. It accepts multimodal input and supports a context window of up to one million tokens, a combination that points squarely at long documents, cross-file lookup and multi-step automation.

Tencent is not shipping the model as a standalone API. Hy4 flash is available inside WorkBuddy and Yuanbao as well as through Tencent Cloud's TokenHub, with free access in WorkBuddy through November 30 and API calls on TokenHub at half the list price. That distribution-first approach mirrors how Chinese vendors typically seed a new model before turning on full pricing.

The more consequential detail is where Hy4 flash sits in the Hunyuan line. It is the family's first flash-tier model, meaning Tencent is now explicitly splitting its lineup into a flagship tier for hard reasoning and a cost-performance tier for high-volume, low-value-per-call traffic. The gap between those two tiers is itself a product strategy.

The economics matter most for agent products. When a task requires reading long documents, looping through tools and self-checking results, token consumption multiplies, and the output-side price often decides whether an automated workflow pays for itself. Cheap output tokens plus a million-token window turn running a model several extra times from a luxury into a default.

Zoom out and the release fits a broader shift: Chinese labs have spent the past year competing less on benchmark peaks and more on inference cost, with light and flash tiers becoming the most crowded part of the lineup. Tencent is filling that slot to defend the call volume flowing through its office and cloud products, and to stay visible in a developer market that is acutely price-sensitive.

Three things are worth watching. Tencent says it plans to release the model's weights and list Hy4 flash on OpenRouter, which would change the cost calculus for third-party hosting and self-deployment. Independent benchmarks will test the claim of near-peer quality at the lowest cost, and post-November usage data will show whether the free window converts into real paid volume.

Why it matters

Tencent has carved out an explicit flash tier, pushing a 224B model into price-per-task territory that suits high-frequency office and agent workloads. If the promised weight release and OpenRouter listing land, China's price-led model competition moves further into the mid-size bracket.

TencentHunyuanModel Release
Back to realtime news

Nearby Updates

All

11/16, 14:33

Zoho rebuilds Campaigns with unified email, SMS, WhatsApp and AI agent workflows

Zoho has rebuilt its Campaigns marketing platform, unifying email, SMS and WhatsApp in one workflow and adding native WhatsApp campaigns plus a new email editor. The platform supports the Model Context Protocol, so compatible AI agents can prepare campaigns, manage audience segments and review performance from natural-language instructions.

11/16, 12:54

Weimob becomes authorized partner for Tencent's Workbuddy and Codebuddy

Weimob has become an officially authorized partner for Tencent's Workbuddy and Codebuddy AI products, according to a Sina Finance report. The announcement ties Tencent's agent tools to Weimob's merchant-services base, though the scope of the authorization, revenue terms and delivery commitments have not been disclosed.

11/16, 12:52

TSMC plans to invest up to $1B in Anthropic and OpenAI

TSMC plans to invest up to $1 billion in Anthropic and OpenAI, according to a Semiecosystem report. The reported move would tie leading-edge chip capacity more closely to the two frontier model developers, though the investment structure, split and confirmation status remain undisclosed.

11/16, 10:52

Alibaba's Qwen4-Max debuts with 2.4T total parameters and 71B active, halving token use

Alibaba's Qwen team has released Qwen4-Max, a model with 2.4 trillion total parameters and 71B active per inference pass, with reports saying it consumes about half the tokens of its predecessor. Coverage also claims a 62% lower cost per task than Qwen3.8-Max while outperforming it.