RunsLocalGPU checkerLocal AI toolsv1.8 · data Oct 14, 2026

All models / Gemma 4 31B

What GPU do I need to run Gemma 4 31B?

Gemma 4 31B has 30.7B parameters. At Q4_K_M with 8K context it needs about 19.8 GB of memory. The smallest graphics card that runs it fully on the GPU is about 24 GB, such as the RTX 3090.

Check your own hardware

Memory by quantization (8K context)

QuantBits per weightMemory needed
Q8_08.533.8 GB
Q6_K6.5626.4 GB
Q5_K_M5.6923.0 GB
Q4_K_M4.8519.8 GB
Q3_K_M3.9116.2 GB
IQ2_M2.711.6 GB

Hardware that runs it fully in fast memory

HardwareMemoryBest quantEst. speed
RTX 409024 GBQ5_K_M28 tok/s
RTX 3090 Ti24 GBQ5_K_M28 tok/s
RX 7900 XTX24 GBQ5_K_M26 tok/s
RTX 309024 GBQ5_K_M26 tok/s
RTX 509032 GBQ6_K43 tok/s
M5 32GB32 GBQ5_K_M4.2 tok/s
M4 32GB32 GBQ5_K_M3.3 tok/s
M4 Max 36GB36 GBQ5_K_M11 tok/s
M4 Pro 48GB48 GBQ8_05.0 tok/s
M4 Max 64GB64 GBQ8_010 tok/s
Ryzen AI Max+ 395 64GB64 GBQ8_04.7 tok/s
RTX PRO 6000 Blackwell96 GBQ8_033 tok/s
M3 Ultra 96GB96 GBQ8_015 tok/s
M4 Max 128GB128 GBQ8_010 tok/s
DGX Spark 128GB128 GBQ8_05.0 tok/s
Ryzen AI Max+ 395 128GB128 GBQ8_04.7 tok/s
M3 Ultra 256GB256 GBQ8_015 tok/s
M3 Ultra 512GB512 GBQ8_015 tok/s

Runs with offload to system RAM (slower)

RTX 5080 ~8.1 tok/s · RTX 5070 Ti ~8.0 tok/s · RTX 4080 Super ~7.6 tok/s · RTX 4080 ~7.6 tok/s · RTX 4070 Ti Super ~7.4 tok/s · RX 9070 XT ~7.3 tok/s · RX 9070 ~7.3 tok/s · RX 7800 XT ~7.3 tok/s · Arc A770 16GB ~7.1 tok/s · RX 6800 XT ~6.9 tok/s · RTX 5060 Ti 16GB ~6.6 tok/s · RX 9060 XT 16GB ~5.7 tok/s · RTX 4060 Ti 16GB ~5.5 tok/s · RTX 3080 Ti ~4.8 tok/s · RTX 3080 12GB ~4.8 tok/s · RTX 5070 ~4.7 tok/s · RTX 4070 Ti ~4.5 tok/s · RTX 4070 Super ~4.5 tok/s · RTX 4070 ~4.5 tok/s · Arc B580 ~4.4 tok/s · RX 7700 XT ~4.4 tok/s · RTX 2080 Ti ~4.2 tok/s · RTX 3060 12GB ~4.2 tok/s · RTX 3080 10GB ~4.0 tok/s

Assumes 32 GB of DDR5 system RAM for offload. Estimates only; real speeds usually land within about 20%.