# Will MOSS 3(in the future) become another Flux 3 / SD3 situation? — retained record

> Retained reddit record from reddit:StableDiffusion.

Canonical: https://highsignal.app/data/records/1c94e7724ee8b430

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:StableDiffusion)  published 2026-10-19 03:44 ingested 2026-10-20 02:38

[original source ↗](https://www.reddit.com/r/StableDiffusion/comments/1x9mlyg/will_moss_3in_the_future_become_another_flux_3/)

## retained text



```
I hope not, but I think we're seeing a potentially disastrous pattern here.

A small, brave startup isn't afraid to release a really good, uncensored model. (Keep in mind, just because you see a guy using an uncensored model doesn't mean he's generating a\*ult stuff... Okay, I did generate a\*ult stuff, but not because the model was uncensored. 😂)

Then they grow bigger and become profitable. Eventually, they release a , censored model. I can understand why companies make that decision, but I'm worried this could become the standard.

In fact, the models I've been happiest with so far are ANIMA, Krea 2, and now MOSS (I only tried it yesterday... :)). All of them come from either a single developer (ANIMA) or a small team(MOSS) or company(Krea 2).

However, from what I've seen, small teams simply don't have enough money to develop good image-editing models and we still wait for the big company like Qwen or BLF.

Honestly, I think Krea 2 has taken a pretty good approach. They used  a\*ult images in training, but didn't go overboard with it, and they use filters to block unwanted content instead. 
```



document fetched 2026-10-20 00:32 · canonical https://reddit.com/r/StableDiffusion/comments/1x9mlyg/will_moss_3in_the_future_become_another_flux_3
