TokenAssemble

Google · open-weights model

gemma-3-12b-it

gemma-3-12b-it is a 12.2B-parameter model by Google. Its smallest quant (Q4_K_M) needs at least 11.0GB of GPU-addressable memory at 8K context; it loads cleanly at a quality quant on 36 of the 50 machines we track.

Parameters
12.2B
Layers
48
Max context
128K
Attention
HYBRID
License
gemma

Config from google/gemma-3-12b-it

Quants and what they need

Measured file sizes; “needs” = weights + KV cache at 8K context + overhead.

QuantFile sizeNeeds ≥
Q8_012.5 GB16.2 GB
Q6_K9.7 GB13.4 GB
Q5_K_M8.4 GB12.1 GB
Q4_K_M7.3 GB11.0 GB

What runs gemma-3-12b-it

Every machine we track, graded at its best clean-fit quant.

HardwareBest-quant verdict · llama.cpp · 8K context
Intel Arc A770 16 GBFits · B · Q6_Kinteractivebeta
Intel Arc B580Tight · C · Q4_K_Minteractivebeta
AMD Radeon PRO W7900Fits · A+ · Q8_0interactivebeta
NVIDIA GeForce RTX 3060 12 GBTight · C · Q4_K_Minteractivebeta
NVIDIA GeForce RTX 3070Tight · C · Q4_K_Minteractive · offloadbeta
NVIDIA GeForce RTX 3070 TiTight · C · Q4_K_Minteractive · offloadbeta
NVIDIA GeForce RTX 3080 10 GBTight · C · Q4_K_Minteractive · offloadbeta
NVIDIA GeForce RTX 3080 12 GBTight · C · Q4_K_Minteractivebeta
NVIDIA GeForce RTX 3080 TiTight · C · Q4_K_Minteractivebeta
NVIDIA GeForce RTX 3090Fits · A+ · Q8_0interactivebeta
NVIDIA GeForce RTX 3090 TiFits · A+ · Q8_0interactivebeta
NVIDIA GeForce RTX 4060Tight · C · Q4_K_Musable · offloadbeta
NVIDIA GeForce RTX 4060 Ti 16 GBFits · B · Q6_Kinteractivebeta
NVIDIA GeForce RTX 4060 Ti 8 GBTight · C · Q4_K_Musable · offloadbeta
NVIDIA GeForce RTX 4070Tight · C · Q4_K_Minteractivebeta
NVIDIA GeForce RTX 4070 SUPERTight · C · Q4_K_Minteractivebeta
NVIDIA GeForce RTX 4070 TiTight · C · Q4_K_Minteractivebeta
NVIDIA GeForce RTX 4070 Ti SUPERFits · B · Q6_Kinteractivebeta
NVIDIA GeForce RTX 4080Fits · B · Q6_Kinteractivebeta
NVIDIA GeForce RTX 4080 SUPERFits · B · Q6_Kinteractivebeta
NVIDIA GeForce RTX 4090Fits · A+ · Q8_0interactivebeta
NVIDIA RTX 5000 Ada GenerationFits · A+ · Q8_0interactivebeta
NVIDIA GeForce RTX 5060 Ti 16 GBFits · B · Q6_Kinteractivebeta
NVIDIA GeForce RTX 5070Tight · C · Q4_K_Minteractivebeta
NVIDIA GeForce RTX 5070 TiFits · B · Q6_Kinteractivebeta
NVIDIA GeForce RTX 5080Fits · B · Q6_Kinteractivebeta
NVIDIA GeForce RTX 5090Fits · A+ · Q8_0interactivebeta
NVIDIA RTX 6000 Ada GenerationFits · A+ · Q8_0interactivebeta
NVIDIA RTX A6000Fits · A+ · Q8_0interactivebeta
AMD Radeon RX 6800Fits · B · Q6_Kinteractivebeta
AMD Radeon RX 6900 XTFits · B · Q6_Kinteractivebeta
AMD Radeon RX 7800 XTFits · B · Q6_Kinteractivebeta
AMD Radeon RX 7900 XTFits · B · Q8_0interactivebeta
AMD Radeon RX 7900 XTXFits · A+ · Q8_0interactivebeta
AMD Radeon RX 9070Fits · B · Q6_Kinteractivebeta
AMD Radeon RX 9070 XTFits · B · Q6_Kinteractivebeta
ASUS Ascent GX10Fits · A+ · Q8_0usablebeta
Beelink GTR9 Pro Ryzen AI Max+ 395 128GBFits · A+ · Q8_0usablebeta
NVIDIA DGX SparkFits · A+ · Q8_0usablebeta
Framework Desktop Ryzen AI Max+ 395 128GBFits · A+ · Q8_0usablebeta
Framework Desktop Ryzen AI Max+ 395 64GBFits · A+ · Q8_0usablebeta
GMKtec EVO-X2 Ryzen AI Max+ 395 128GBFits · A+ · Q8_0usablebeta
HP Z2 Mini G1a Ryzen AI Max+ PRO 395 128GBFits · A+ · Q8_0usablebeta
Mac mini M4 16GBTight · C · Q4_K_Musablebeta
Mac mini M4 24GBFits · B · Q6_Kpainfulbeta
Mac mini M4 Pro 24GBFits · B · Q6_Kusablebeta
Mac mini M4 Pro 48GBFits · A+ · Q8_0usablebeta
Mac Studio M3 Ultra 96GBFits · A+ · Q8_0interactivebeta
Mac Studio M4 Max 36GBFits · A+ · Q8_0usablebeta
Mac Studio M4 Max 64GBFits · A+ · Q8_0interactivebeta

Verdicts computed from published specs and measured GGUF sizes — how the math works.