Tencent launched Hy4 Flash on Monday, November 16 — a cheaper, faster sibling of its Hy4 flagship built for everyday office work like research, slide decks, documents and spreadsheets. Why it matters: Tencent is pricing it at a fraction of flagship rates and giving it away free in its WorkBuddy app through November 30, a direct push to make AI a default tool for routine work rather than a premium extra.
November 16, 2026
What Tencent shipped
Hy4 Flash is a multimodal model with 224 billion total parameters, of which about 12 billion are active on each forward pass, and a context window of up to one million tokens. It accepts image as well as text input and inherits the agentic abilities of the full Hy4 — Tencent positions it for deep search, document and spreadsheet editing, presentation building and operating software on a user’s behalf.
The model is live now on Tencent Cloud’s TokenHub under the model ID hy4-flash, and inside Tencent’s WorkBuddy and Yuanbao apps. Tencent says an OpenRouter listing and an open-weight release of the model are both planned, though it has given no date for the weights.
The pricing is the story
List prices on TokenHub are $0.07 per million input tokens, $0.281 per million output tokens and $0.003 per million cached input tokens. Through November 30, the API is half price, and WorkBuddy users get the model free. On Tencent’s internal WorkIndex benchmark, the company says Hy4 Flash completed everyday tasks at one-twentieth of the per-task cost of the full Hy4 flagship.
Tencent also ran a blind test in which 85 WorkBuddy users rated outputs across 259 everyday work tasks. Hy4 Flash scored 2.79 out of 4 on user satisfaction — effectively level with Anthropic’s Claude Haiku 5.5 at 2.81 and ahead of GLM-5.3-Flash at 2.53 — while posting the lowest cost and shortest completion time of the models tested. Those are Tencent’s own numbers from its own test, not independent evaluations, and they should be read that way.
Why it matters
The flagship race gets the headlines, but the volume in AI is in cheap, good-enough models doing unglamorous work all day. Hy4 Flash is Tencent’s bet that a model a twentieth of the flagship’s running cost, bundled free into an office app, wins more real usage than a benchmark crown. For subscribers, the practical takeaway is simple: if you use Tencent’s tools, capable AI assistance just got effectively free for the next two weeks — and the price war among Chinese labs keeps pushing the cost of everyday AI work toward zero. Everything AI, tested: we will treat Tencent’s internal cost and satisfaction claims as vendor claims until independent testing catches up.
Frequently asked questions
How much does Hy4 Flash cost?
List price is $0.07 per million input tokens and $0.281 per million output tokens on Tencent Cloud TokenHub. The API is half price until November 30, 2026, and the model is free in Tencent’s WorkBuddy app over the same period.
Is Hy4 Flash open source?
Not yet. Tencent says it plans to release the weights as an open-weight model, but it has not announced a date.
How does Hy4 Flash differ from the full Hy4?
It is much smaller — 224 billion total parameters with 12 billion active, versus 770 billion and 49 billion for the flagship — and far cheaper per task, while keeping multimodal input, a one-million-token context window and agentic features.
Sources: TechNode, KuCoin (ChainCatcher), The9

