So I tried cramming FLUX into under 1GB, is Bonsai's quantization actually worth it?

Just saw PrismML did extreme quantization on FLUX.2 Klein 4B, 1-bit can squeeze it down to 0.93GB, ternary version is 1.21GB, Apache 2.0 license. They’re claiming it’s the first 4B-level image model that can run on a phone. What I care about is the trade-offs behind the numbers: the original FP16 transformer needs 7.

75GB, around 13GB VRAM, a 4070 can run it but phones are a no-go; after quantization, the size gets cut to one-sixth. Quality-wise, they admit it inherits the ceiling of the Klein distilled version, and 1-bit shows visible degradation to the naked eye. So my take is—if you’ve got a dedicated GPU on desktop, go for the ternary version, 90% quality at one-sixth the size, best bang for your buck; 1-bit only makes sense on phones or devices with under 2GB VRAM.

This is a win for portability, not a breakthrough in image quality, don’t expect it to look better than the original.

That 95% quality with three values is pretty impressive.

Running 1bit on a phone is more of a novelty than actually useful.

“Portability wins” — that’s a spot-on way to put it.

The distillation ceiling is holding everything back, no matter how aggressive you get with quantization, you can’t push past it.

0.93GB is actually insane, and it can still generate images.

If your 4070 can handle it, no need to mess with 1bit.

Saved the post.