Models › gemma-4-12B-it

gemma-4-12B-it: VRAM needs and speed by GPU

At Q4_K_M, gemma-4-12B-it needs about 8.2 GB of VRAM. No measured runs yet. Speeds below are estimates. Submit a run.

11.91B parameters · license: apache-2.0 · 262,144-token context · GGUF published May 29, 2026 · GGUF files on Hugging Face

Memory needed per quantization

QuantFile sizeNeeds (estimate)Smallest card it fits
UD-IQ2_M4.2 GB5.3 GB8 GB cards
UD-IQ3_XXS4.6 GB5.7 GB8 GB cards
UD-Q2_K_XL4.7 GB5.8 GB8 GB cards
Q3_K_S5.1 GB6.2 GB8 GB cards
Q3_K_M5.7 GB6.8 GB8 GB cards
UD-Q3_K_XL6.0 GB7.1 GB8 GB cards
IQ4_XS6.4 GB7.5 GB8 GB cards
IQ4_NL6.7 GB7.8 GB12 GB cards
Q4_06.7 GB7.8 GB12 GB cards
Q4_K_S6.8 GB7.9 GB12 GB cards
Q4_K_M7.1 GB8.2 GB12 GB cards
UD-Q4_K_XL7.4 GB8.5 GB12 GB cards
Q4_17.4 GB8.5 GB12 GB cards
Q5_K_S8.2 GB9.3 GB12 GB cards
Q5_K_M8.4 GB9.5 GB12 GB cards
UD-Q5_K_XL8.6 GB9.7 GB12 GB cards
Q6_K9.8 GB10.9 GB12 GB cards
UD-Q6_K_XL10.7 GB11.8 GB16 GB cards
Q8_012.7 GB13.8 GB16 GB cards
UD-Q8_K_XL13.6 GB14.7 GB16 GB cards
BF1623.8 GB24.9 GB32 GB cards

Expected speed on popular hardware

HardwareBest quant that fitsExpected tok/s
NVIDIA RTX 3060Q6_K31.5 estimate
NVIDIA RTX 4060 TiIQ4_XS38.4 estimate
NVIDIA RTX 3090Q8_051.8 estimate
NVIDIA RTX 4090Q8_057.4 estimate
NVIDIA RTX 5090Q8_090.6 estimate
AMD Radeon RX 7900 XTXQ8_057.6 estimate
Apple M4 Pro (20-core GPU) (64 GB)Q8_018.3 estimate
Apple M3 Max (40-core GPU) (64 GB)Q8_026.7 estimate
AMD Ryzen AI Max+ 395 (Radeon 8060S)Q8_016.4 estimate

Check your own card

Too big for your card?

Run gemma-4-12B-it on a rented 24-80 GB GPU by the hour.

Rent a bigger GPU on RunPod

Affiliate link: if you sign up, RigBench may earn a referral credit. It never affects rankings or numbers.