community pool / model answer

gemma-4-E2B-it

chat · google/gemma-4-E2B-it

107 tok/s · model median · 15 runs in the model record.

Reference rig	Quantization	Median result	Basis	TTFT (ms)	Peak VRAM	Max context	Engines
AMD Ryzen 7 8745H w/ Radeon 780M	4-bit	47.4 tok/s	reported	485	2.5 GB	32,768	llama.cpp
Intel(R) Core(TM) Ultra 7 155H	4-bit	27.6 tok/s	reported	-	- GB	2,048	llama.cpp
Qualcomm Snapdragon 888 ARM64	4-bit	6.2 tok/s	reported	10,000	6 GB	1,306	llama.cpp
GTX 1060 3GB	4-bit	48.2 tok/s	reported	8,399	- GB	2,048	llama.cpp
Radeon AI Pro R9700 32GB	8-bit	169.7 tok/s	reported	-	- GB	2,048	llama.cpp
Radeon AI Pro R9700 57GB	4-bit	47.4 tok/s	reported	493	- GB	2,048	llama.cpp
RTX 3060 12GB	4-bit	133.4 tok/s	reported	112	3.8 GB	4,096	llama.cpp
RX 570 4GB ×2	2-bit	16 tok/s	reported	246	- GB	2,048	llama.cpp
RX 5700 8GB	4-bit	73.1 tok/s	reported	-	3 GB	2,048	llama.cpp
Ryzen AI Max 395 128GB	4-bit	107 tok/s	measured	-	8.5 GB	32,768	llama.cpp
Ultra 7 155H 32GB	4-bit	28 tok/s	reported	-	- GB	2,048	llama.cpp

Context tested

The table above is the tested context for each exact rig and quantization cell. Empty fields remain empty rather than being filled with an estimate.