Been working on Agent Mode 👀
The goal is simple: let the agent work directly with the tools you already use, without disrupting your existing workflow.
Here’s a little demo 🐈
It's impressive how fast Chrome Web Store reviews have become!
Less than an hour, and an update is already reviewed and published..
In the past, this could take 3 days at best, or even a week.
Kudos to @ChromiumDev! 👏
ChatGPT made some changes on their backend that caused every Chrome extension side panel to crash :)
We’ve fixed it in the latest version, so update now!
Also, for BYOK chat users, Agent mode is cooking 👀
Ask your AI assistant to create a game for you while you're browsing or working, and in your free time...
Soon it will be available in BYOK and local models..
You built it, now show it off and share it.
With shareable profiles, you can bring your Sites and plugins together in one place so others can find them, and reuse them. Teammates can discover your shared skills, too—those stay within your workspace.
Available to Free, Go, Plus, Pro, and Business users. Rolling out to ChatGPT Enterprise, Edu, and Healthcare plans soon.
My AI questions start on other pages. Then: new tab, paste, wait, come back.
So I built Quick chat into chat.cx.
Select text anywhere, ask, and the answer opens beside it.
Uses your own ChatGPT, Claude or Gemini accounts.
Live now on the Chrome Web Store.
🥷 New stealth model: Union Alpha is now free on OpenRouter.
You can use it directly in chat.cx, just add your free OpenRouter API key.
One panel. Every AI. 👀
chat.cx
Chrome now has a local AI model running directly in your browser. 🤯
And yes — you can disconnect from the internet and still chat with it.
Here's what I built with it 🧵👇
🧪 Gemma 4
Chrome 153 also includes a Gemma 4 for Built-in AI flag:
chrome://flags/#gemma4-for-built-in-ai
Enable it, restart Chrome, and you can use Gemma 4 through the same Built-in AI APIs.
NVIDIA just added deepseek-v4-pro-0813 to NIM.
Free endpoint API.
1.6T params. 1M context. #2 on SWE-bench Verified.
30 seconds to wire it into chat.cx 👇
Name → nvidia
URL → integrate.api.nvidia.com/v1
Model → deepseek-ai/deepseek-v4-pro-0813