zev

imagination amplified

[zev @zevoo]

Flux 3 for finding an idea.

[zev @zevoo]

Flux 3 is beautiful

[zev @zevoo]

More Flux 3

[zev @zevoo]

Flux 3 can do the old videogame style pretty well!

[zev @zevoo]

Fun with flux 3

[zev @zevoo]

Flux 3 can do backflips pretty well!

[zev @zevoo]

Made with Flux 3, what a fun model.

[zev @zevoo]

Having fun with Flux 3, thanks! @bfl_ai @NousResearch

[zev @zevoo]

@1DeGhost Would love to your rigging process. Looks great!

[zev @zevoo]

@VictorTaelin Hype it up by building something with it that would be extremely challenging without it. Show us its usefulness. Excited for your progress.

[zev @zevoo]

Please Anthropic keep Fable on subscription. It's just unlike anything before it.

[zev @zevoo]

@ostrisai Would be nice to have a dedicated sampling page to test out loras in the app.

[zev @zevoo]

@fofrAI I think sora could have been great, the problem isn't even copyright restrictions. They actually just gutted the model quality a few weeks after release to make room for more people. It's not the same model at all.

[zev @zevoo]

@gabriel1 @vparv3 The real problem is that the model was nerfed shortly after launch. Objectively worse with the same prompts. Probably to make way for more people but the current sora 2 model is much worse than the launch model.

[zev @zevoo]

@hive_echo You've been on a roll with making really cool stuff lately. Keep it up!

[zev @zevoo]

the argument for scaling compute for potential AGI, though scaling compute alone might not lead to AGI the models that are born from it could help push the breakthroughs that will.

[zev @zevoo]

Nano bannana pro is really good

[zev @zevoo]

@nearcyan @omooretweets The real problem is that sora at launch was a much stronger model and was nerfed shortly after more people got let in. The model week 1 was really really good. Really sad

[zev @zevoo]

Can we get gemini 3 today please and thank you

[zev @zevoo]

@ericzakariasson Must be nice but sadly that would use my entire monthly usage in one session