Just saw some news — Zhixiang Future's image gen model ranked #2 globally now.

资讯 / AI 快讯
1

So apparently ZhiXiang Future’s image gen model hit #2 globally on some leaderboard, only behind OpenAI. Honestly kinda surprised a domestic model managed to squeeze into that spot.

But rankings have never really been my thing—they didn’t even specify which leaderboard this is or what dimensions they’re scoring on. The real test is just throwing a few images in and seeing how it performs. Rankings alone don’t mean much. Anyone here already tried it out? How’s the actual output quality? Especially curious about the two classic pain points: Chinese semantic understanding and hands.

2

Being #2 on the leaderboard doesn’t mean it’s actually good in practice—you gotta see how it holds up in real-world scenarios.

3

It’s good to see domestic options catching up — more choices for everyone.

4

Which review board are you even talking about? Just saying “second place” sounds kinda sketchy.

5

Honestly, hands and Chinese text layout are the real test of a model.

6

OpenAI being #1 isn’t really surprising.

7

Waiting for a side-by-side comparison. Gotta test it yourself to know for sure.

8

Chinese typography has always been a weak spot for domestic models, so making it into the top two means they actually put some serious work into it. Just looking at the numbers, though, it does feel a bit hollow.

9

Hands and Chinese text layout are the real acid test, whether it ranks #2 or not doesn’t mean much