Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

Gemini 4 Hydrogen hands-on review shows stronger self-verification but uneven reliability

AI summary

A Juejin author reports that Google made Gemini 4 Hydrogen available to Google AI Pro users and tested it on web-page replication, game creation and controller interaction. The author found that it often used a browser, screenshots and simulated clicks to verify results, while producing cleaner pages with the same prompts. Reliability was inconsistent: a search during a Super Mario test ran for 30 minutes without finishing, and another task received only basic syntax validation rather than screenshot and interaction testing. The article also reproduces Google's benchmark claims and the author's theoretical Gemini 4 model ordering, while noting that external rankings conflict.

In-depth insight

The hands-on findings indicate that Hydrogen can use agent tools for more comprehensive self-verification than earlier Google models, but may not apply those tools consistently in engineering tasks.

最近谷歌好像吃错药了一样,突然变猛男了?

模型抛了一个又一个。

抛出 Argon 之后,接着又上了 Carbon,没几天又发了 Hydrogen,听说后面还有一个更强的 Krypton。

我本来还想吐槽一下谷歌的,Carbon 居然只给 Ultra 会员用。现在终于给 Pro 来了个 Hydrogen,可以测起来。

就做两件事情:

#Google#Gemini 4 Hydrogen#Google AI Pro

Original · 10 min readRead the original
Group chat

No comments yet. Start the conversation.