Configs › gemma-4-E2B-it on RTX 5090: 340.6 tok/s measured

gemma-4-E2B-it on RTX 5090: 340.6 tok/s measured

Measured gemma-4-E2B-it speeds: 340.6 tok/s on RTX 5090 (Q4_0, llama.cpp/cuda). 19 configs, each linked to its source, snapshot 2026-10-15.

Snapshot 2026-10-15 · 394 configs on AI boxes and 423 on GPU cards · 123 models. Every number is generated from the tracker snapshot and linked to where it was measured; nothing here is typed by hand.

Quick answers

How fast is gemma-4-E2B-it on RTX 5090?: 340.6 tok/s single-stream decode (Q4_0, llama.cpp/cuda), the fastest of 19 measured configs on NVIDIA RTX 5090 (32GB), measured by paulplee, source last checked 2026-10-15. source

gemma-4-E2B-it on GPU cards, ranked on their own

GPU cards run the model from their own VRAM: very fast on what fits, but most cannot load what a 128 GB AI box can. Ranked on their own, never against the boxes.

Model	Hardware	Quant	Backend	Decode	Context	Trust	Source
gemma-4-E2B-it	RTX 5090	Q4_0	llama.cpp/cuda	340.6 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	IQ4_XS	llama.cpp/cuda	338.2 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	IQ4_NL	llama.cpp/cuda	337.3 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	Q4_1	llama.cpp/cuda	332.1 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	UD-IQ3_XXS	llama.cpp/cuda	331.9 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	Q4_K_S	llama.cpp/cuda	323.1 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	UD-Q3_K_XL	llama.cpp/cuda	318.5 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	Q4_K_M	llama.cpp/cuda	317.5 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	UD-Q4_K_XL	llama.cpp/cuda	317.3 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	Q3_K_M	llama.cpp/cuda	316.7 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	Q5_K_S	llama.cpp/cuda	310.8 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	Q3_K_S	llama.cpp/cuda	310.1 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	Q5_K_M	llama.cpp/cuda	308.4 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	UD-Q5_K_XL	llama.cpp/cuda	303.3 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	Q6_K	llama.cpp/cuda	296.4 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	UD-Q6_K_XL	llama.cpp/cuda	285.5 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	Q8_0	llama.cpp/cuda	278.3 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	UD-Q8_K_XL	llama.cpp/cuda	271.3 tok/s	8192	Structured table	source
gemma-4-E2B-it	RTX 5090	BF16	llama.cpp/cuda	214.2 tok/s	8192	Structured table	source

More: all configs · hardware · news · contributors · methodology · llms.txt · llms-full.txt · API (OpenAPI) · JSON snapshot. Built by Altronis, private on-prem AI, Singapore.