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

All models / Gemma 4 E2B

What GPU do I need to run Gemma 4 E2B?

Gemma 4 E2B has 5.1B parameters (2.3B active per token, mixture-of-experts). At Q4_K_M with 8K context it needs about 3.8 GB of memory. The smallest graphics card that runs it fully on the GPU is about 8 GB, such as the RTX 4060.

Check your own hardware

Memory by quantization (8K context)

QuantBits per weightMemory needed
Q8_08.56.1 GB
Q6_K6.564.9 GB
Q5_K_M5.694.3 GB
Q4_K_M4.853.8 GB
Q3_K_M3.913.2 GB
IQ2_M2.72.4 GB

Hardware that runs it fully in fast memory

HardwareMemoryBest quantEst. speed
RTX 5060 Ti 8GB8 GBQ8_0110 tok/s
RTX 50608 GBQ8_0110 tok/s
RTX 30708 GBQ8_0110 tok/s
RTX 3060 Ti8 GBQ8_0110 tok/s
RTX 50508 GBQ8_079 tok/s
RTX 4060 Ti 8GB8 GBQ8_071 tok/s
RX 76008 GBQ8_071 tok/s
RTX 40608 GBQ8_067 tok/s
RTX 3080 10GB10 GBQ8_0150+ tok/s
RTX 2080 Ti11 GBQ8_0150+ tok/s
RTX 3080 Ti12 GBQ8_0150+ tok/s
RTX 3080 12GB12 GBQ8_0150+ tok/s
RTX 507012 GBQ8_0150+ tok/s
RTX 4070 Ti12 GBQ8_0124 tok/s
RTX 4070 Super12 GBQ8_0124 tok/s
RTX 407012 GBQ8_0124 tok/s
Arc B58012 GBQ8_0112 tok/s
RX 7700 XT12 GBQ8_0106 tok/s
RTX 3060 12GB12 GBQ8_088 tok/s
RTX 508016 GBQ8_0150+ tok/s
RTX 5070 Ti16 GBQ8_0150+ tok/s
RTX 4080 Super16 GBQ8_0150+ tok/s
RTX 408016 GBQ8_0150+ tok/s
RTX 4070 Ti Super16 GBQ8_0150+ tok/s
RX 9070 XT16 GBQ8_0150+ tok/s
RX 907016 GBQ8_0150+ tok/s
RX 7800 XT16 GBQ8_0150+ tok/s
Arc A770 16GB16 GBQ8_0137 tok/s
RX 6800 XT16 GBQ8_0126 tok/s
RTX 5060 Ti 16GB16 GBQ8_0110 tok/s
RX 9060 XT 16GB16 GBQ8_079 tok/s
RTX 4060 Ti 16GB16 GBQ8_071 tok/s
M5 16GB16 GBQ8_038 tok/s
M4 16GB16 GBQ8_029 tok/s
RX 7900 XT20 GBQ8_0150+ tok/s
RTX 409024 GBQ8_0150+ tok/s
RTX 3090 Ti24 GBQ8_0150+ tok/s
RX 7900 XTX24 GBQ8_0150+ tok/s
RTX 309024 GBQ8_0150+ tok/s
M4 Pro 24GB24 GBQ8_067 tok/s
RTX 509032 GBQ8_0150+ tok/s
M5 32GB32 GBQ8_038 tok/s
M4 32GB32 GBQ8_029 tok/s
M4 Max 36GB36 GBQ8_0101 tok/s
M4 Pro 48GB48 GBQ8_067 tok/s
M4 Max 64GB64 GBQ8_0134 tok/s
Ryzen AI Max+ 395 64GB64 GBQ8_063 tok/s
RTX PRO 6000 Blackwell96 GBQ8_0150+ tok/s
M3 Ultra 96GB96 GBQ8_0150+ tok/s
M4 Max 128GB128 GBQ8_0134 tok/s
DGX Spark 128GB128 GBQ8_067 tok/s
Ryzen AI Max+ 395 128GB128 GBQ8_063 tok/s
M3 Ultra 256GB256 GBQ8_0150+ tok/s
M3 Ultra 512GB512 GBQ8_0150+ tok/s

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