Request: Gemma 4 8B/12B

[WizzrdLord]

Just tested Gemma 4 E4B, it is amazing for a model its size. However there are some things that a model its size still cannot do.
Gemma 4 26B is probably amazing, I haven't gotten the chance to run it locally because of my 16GB VRAM limitation, which brings me to what I wanted to discuss with you all.
Do you think Google should come out with a model that is somewhere between 4B and 26B? Maybe a 8 or 12B model?
I feel like a 22B parameter gap is too big.

The 22-billion parameter jump creates a "no-man’s-land" for developers on standard hardware. A 16GB MacBook (the industry standard for mobile development) can run E4B with ease, but the 26B MoE is way too big for it.