Loading video...

Video Failed to Load

Go Home

31,842 views • 19 hours ago •via X (Twitter)

21 Comments

ChinaTechFlux's profile picture
ChinaTechFlux19 hours ago

Pelican bike test on hy4. Pure SVG, no image model. Diamond frame, 12 spokes, throat pouch, webbed feet on the pedals. Animated: 12-phase legs, 2:1 gearing, 1.6s loop, zero pixel drift at the seam. Genuinely good.

Canopy Wave's profile picture
Canopy Wave19 hours ago

Hello, Hy4. Looking good already

PokeBench's profile picture
PokeBench19 hours ago

Great work. Impressive improvement from the preview

Axel ker's profile picture
Axel ker19 hours ago

Tried it out on @PerchTerminal it’s really fast. Excited to see if it actually holds up against Ds flash

Pandaily's profile picture
Pandaily19 hours ago

Two numbers stand out in Tencent's release: search hallucinations down from 15.76% on the preview to 5.45%, and API pricing held at the preview's $0.834 per million input tokens despite the new vision and computer-use skills.

Michael Waitze's profile picture
Michael Waitze19 hours ago

It is impressive to see the vision and computer use capabilities in this release.

Zidan Silverlane's profile picture
Zidan Silverlane14 hours ago

Bravo 👏

DoN's profile picture
DoN19 hours ago

Hi 4

Adolan's profile picture
Adolan19 hours ago

I downloaded WorkBuddy to test out the model and got this when I tried to sign up for an account. Not a good look!

E1ev3n.'s profile picture
E1ev3n.19 hours ago

I am curious how the new vision holds up on messy real screens, because my own tests usually break on the second window, not the first. I will run it on a cluttered desktop before I trust the demo.

Donut's profile picture
Donut13 hours ago

The launch post that actually ships vision + computer use in one drop is rarer than another benchmark dump. Weights still the line that turns "try the API" into "run it yourself."

Bowen LIU's profile picture
Bowen LIU16 hours ago

感觉比昨晚那个宣传片好看啊

J.W's profile picture
J.W17 hours ago

官博自己留了后路:权重只是计划开源,系统卡也等权重一起发;现在能用的是同价 API,不是本地可下的开源权重。

Daniel Tedesco's profile picture
Daniel Tedesco17 hours ago

yes yes, great, but can it find the frog @aisearchio

Canopy Wave Developers's profile picture
Canopy Wave Developers19 hours ago

“To See” has our attention. 👀👀👀👀

jimerrors's profile picture
jimerrors16 hours ago

Curious how Hy4's vision holds up outside the demo — that 70-second clip makes screenshots look easy, is it that smooth in practice?

AIHubmix's profile picture
AIHubmix17 hours ago

congrats team on release

meguna's profile picture
meguna19 hours ago

@opencode @thdxr Hurry up and Make it free Mother Fuckers.

Rithesh's profile picture
Rithesh16 hours ago

Congrats . tested and created this video

Dextro's profile picture
Dextro15 hours ago

Introducing Avatar the air bender in your screen

leehoo's profile picture
leehoo19 hours ago

Hy4 正式介绍片出来了:视觉和电脑操作一起上。开源这桌现在拼的不是海报谁更炸,是谁真能把金融、前端、办公这些活干完——先拿真实场景测一轮再说。

Related Videos

🚨 I just built a game with an open-source AI model. And honestly… I didn’t expect it to be this capable. Tencent Hunyuan just released Hy4 preview, and it’s already pushing into the top tier of open-source models. Three major releases in six months. That pace is crazy. Here’s what Hy4 preview brings: → 770B total parameters → 49B active parameters → 1M+ token context window → Fully open-source But the numbers aren’t even the most interesting part. Hy4 preview was built around one goal: real-world productivity. Coding. Engineering. Office work. Science. Gaming. Finance. Security. And Tencent didn’t build it in isolation. Hy4 preview was co-designed alongside real products like WorkBuddy, using expertise and real-world data from across Tencent’s ecosystem. So I decided to test it the way I actually like testing AI models: I gave it a game idea and let WorkBuddy help turn it into a playable experience. 🎮 From the initial concept to the actual game logic, it was surprisingly smooth. And the benchmark results back up the hype: 163 internal experts 203 engineering tasks Hy4 preview — 2.99/4 Kimi K3 — 2.94/4 GLM 5.3 — 2.92/4 It also beats GLM 5.2 on benchmarks and comes remarkably close to GLM 5.3. Then comes the part I really like: 💰 ¥6/M input tokens 💰 ¥18/M output tokens 💰 ¥0.30/M cache hits Flagship-level capability without the flagship-level price. And right now, you can try Hy4 preview FREE through WorkBuddy for the next two weeks. If you’re curious what it can actually do, don’t just read the benchmarks. Build something with it. 🔗 Tencent Hy Tencent AI

Aryan Rakib

63,296 views • 1 month ago

Introducing.
0:25

Sensitive content

Introducing.

Useless Utility

70,049 views • 3 years ago