Lei Tech did a full comparison of domestic video models, Kling and Jimeng are pretty much on par.

Lei Tech did a head-to-head review of domestic video models, and the conclusion was that Kling and Jimeng are pretty much on par. They also teased that the worst performer turned out to be some specific one—didn’t name it in the title, so I won’t bother guessing. I occasionally use video models for rough dynamic sketches when I’m drawing, and Kling and Jimeng are definitely the two I find most intuitive.

Each has its own quirks—Kling has smoother motion, while I prefer Jimeng’s visual quality. Hard to say which is absolutely better.

Reviews like this are just for reference anyway; different prompts and scenarios can lead to wildly different conclusions. That “worst” bit though, it’s got me tempted to dig up the original article. Who do you guys think is the worst? Let’s compare answers.

Yeah, Kling’s motion handling is definitely more stable than Jimeng’s.

I think I know which one’s the worst, but I’m not gonna say.

The quality of the prompts used for a side-by-side comparison is way too important.

Clickbait title, OP didn’t even say who the worst was.

Everyone uses them differently, ranking them is pointless.

Got a link to the original post?

I’ve been switching between Kling and Jimeng through the pixpix portal to compare them side by side—run the same material through each, and you can tell which one’s better right away.

Honestly, when it comes to these head-to-head comparisons, the methodology matters way more than the final ranking. If you pick a slightly different prompt, the whole order shifts. So it’s way more useful to look at how they tested it than who came out on top.

1 Like