AI Personal Learning
and practical guidance
讯飞绘镜

Mysterious image generation model Red panda (Red Panda) topped the arena! More than FLUX 1.1Pro

Recently, in the global Artificial Analysis (AI) modeling arena (Artificial Analysis) Vincennes graph model ranking, a model namedRed panda's new model suddenly killed it with a 9 %'s win rate surpassed that of the original #1 big brother Flux1.1Pro Become the new king!

神秘图像生成模型Red panda(红熊猫)登顶竞技场!超过FLUX1.1Pro-1


 

神秘图像生成模型Red panda(红熊猫)登顶竞技场!超过FLUX1.1Pro-1

Red panda updated rankings, still leading 3%

 

As you can see, red_panda has a score of 1,144, and a full 9% more arena wins than Flux~.

This shocked the foreign netizens, what model can actually surpass Flux 1.1 so much?

神秘图像生成模型Red panda(红熊猫)登顶竞技场!超过FLUX1.1Pro-1

 

The Red Panda model is currently not open for port use and can only be blind tested in the arena~

In order to visualize the comparison, we conducted several rounds of testing in the Artificial Analysis arena and took the following comparison charts ↓

Cue word: Remains of ancient civilization found on barren Mars

神秘图像生成模型Red panda(红熊猫)登顶竞技场!超过FLUX1.1Pro-1Left: Red_panda Right: Flux.1 [Dev]

Cue word: There is a high-tech laboratory on Mars with sophisticated instruments and staff working on the

神秘图像生成模型Red panda(红熊猫)登顶竞技场!超过FLUX1.1Pro-1Left: MidjourneyV6.1 Right: Red_panda

Cue word: an elevator from Mars to the moon.

神秘图像生成模型Red panda(红熊猫)登顶竞技场!超过FLUX1.1Pro-1Left: Red_panda Right: Stable diffusion 3

Cue in: the Moon Hotel, where people stand in front of floor-to-ceiling windows and look out the window into the universe

神秘图像生成模型Red panda(红熊猫)登顶竞技场!超过FLUX1.1Pro-1Left: Flux 1.1 Right: Red_panda

After comparing them, one of the most intuitive feelings I get from the Red panda is that - the

Real, too real...

It seems to have completely gotten away from the greasy and unrealistic feel of the AI models, even far beyond the Flux models which are already incredibly realistic!

And a very good understanding of the cue word sentence ↓

Cue word: Time-lapse images of pedestrians walking and vehicles weaving on the streets of a modern metropolis highlight the fast-paced atmosphere and pulse of the city.

神秘图像生成模型Red panda(红熊猫)登顶竞技场!超过FLUX1.1Pro-1Left: Red panda Right: SD3.5 large turbo

More and more, I want to know who this mysterious red panda model is from~

Some foreign netizens speculate that it may be the Midjourney-V7 model

神秘图像生成模型Red panda(红熊猫)登顶竞技场!超过FLUX1.1Pro-1There is even a speculation from the netizens - based on the name of the model "Red Panda", it is suspected that it is a new model trained by a Chinese company.

神秘图像生成模型Red panda(红熊猫)登顶竞技场!超过FLUX1.1Pro-1I have to say that the imagination of this class of netizens is as wonderful as the Red_panda model.

I don't think it's like the models trained by Tencent and Baidu, if it's trained domestically, the first thing to consider is definitely Tsinghua's new framework or Doubao's raw map model.

Most likely Midjourney's V7 model or real based on SD3.5 larg turbo trainingLoraFine-tuning the model.

And there is a higher chance that this "Red_panda" model is the latter.

Because tomorrow, that is.The 29th.It will be launched Medium cup version of SD3.5 I'm sure they're just trying to build up the traffic for it.

Looking forward to it, so which company's model do you think it is?

May not be reproduced without permission:Chief AI Sharing Circle " Mysterious image generation model Red panda (Red Panda) topped the arena! More than FLUX 1.1Pro
en_USEnglish