Models › gemma-3-27b-it

gemma-3-27b-it: VRAM needs and speed by GPU

At Q4_K_M, gemma-3-27b-it needs about 18.0 GB of VRAM. No measured runs yet. Speeds below are estimates. Submit a run.

27.01B parameters · license: gemma · 131,072-token context · GGUF published Mar 12, 2025 · GGUF files on Hugging Face

Memory needed per quantization

QuantFile sizeNeeds (estimate)Smallest card it fits
UD-IQ1_S6.5 GB8.0 GB12 GB cards
UD-IQ1_M7.0 GB8.5 GB12 GB cards
UD-IQ2_XXS7.9 GB9.3 GB12 GB cards
UD-IQ2_M9.6 GB11.1 GB12 GB cards
Q2_K10.5 GB12.0 GB16 GB cards
Q2_K_L10.5 GB12.0 GB16 GB cards
UD-Q2_K_XL10.7 GB12.2 GB16 GB cards
UD-IQ3_XXS10.8 GB12.3 GB16 GB cards
Q3_K_S12.2 GB13.6 GB16 GB cards
Q3_K_M13.4 GB14.9 GB16 GB cards
UD-Q3_K_XL13.7 GB15.2 GB16 GB cards
IQ4_XS14.8 GB16.2 GB24 GB cards
IQ4_NL15.6 GB17.0 GB24 GB cards
Q4_015.6 GB17.1 GB24 GB cards
Q4_K_S15.7 GB17.2 GB24 GB cards
Q4_K_M16.5 GB18.0 GB24 GB cards
UD-Q4_K_XL16.8 GB18.3 GB24 GB cards
Q4_117.2 GB18.6 GB24 GB cards
Q5_K_S18.8 GB20.2 GB24 GB cards
Q5_K_M19.3 GB20.8 GB24 GB cards
UD-Q5_K_XL19.3 GB20.8 GB24 GB cards
Q6_K22.2 GB23.6 GB32 GB cards
UD-Q6_K_XL23.7 GB25.2 GB32 GB cards
Q8_028.7 GB30.2 GB32 GB cards
UD-Q8_K_XL31.8 GB33.3 GB48 GB cards
BF1654.0 GB55.5 GB96 GB cards

Expected speed on popular hardware

HardwareBest quant that fitsExpected tok/s
NVIDIA RTX 3060UD-IQ2_M32.0 estimate
NVIDIA RTX 4060 TiDoes not fit
NVIDIA RTX 3090Q5_K_M34.0 estimate
NVIDIA RTX 4090Q5_K_M37.7 estimate
NVIDIA RTX 5090Q8_040.0 estimate
AMD Radeon RX 7900 XTXQ5_K_M37.9 estimate
Apple M4 Pro (20-core GPU) (64 GB)Q8_08.1 estimate
Apple M3 Max (40-core GPU) (64 GB)Q8_011.8 estimate
AMD Ryzen AI Max+ 395 (Radeon 8060S)Q8_07.2 estimate

Check your own card

Too big for your card?

Run gemma-3-27b-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.