Google

gemma-3n-E4B-it8B

Name: gemma-3n-E4B-it
Author: Google

Relabel 2026-09-23: Laeufe waren faelschlich als gemma-4-E4B-it erfasst (GGUF-Metadata = Gemma 3n E4B)

Dense8B

Position in the field

Best values compared

Best metrics of this model against the minimum, average and maximum of all published systems.

Generation277,5 tok/s

Min 0,0Ø 242,0Max 2.491,2

Ueber dem Durchschnitt · 4261 Systeme im Feld

Prefill4.348 tok/s

Min 4Ø 4.687Max 55.260

Unter dem Durchschnitt · 4261 Systeme im Feld

Time to First Token571 ms

Min 20Ø 32.760Max 535.235

Ueber dem Durchschnitt · 4259 Systeme im Feld

Performance profile

Each bubble is a GPU, CPU or engine – position shows prompt processing (X) and output speed (Y), bubble size the number of runs.

AMD Radeon AI PRO R9700 277,5 tok/s

AMD Ryzen 3 3100 4-Core Processor 277,5 tok/s

llama.cpp 277,5 tok/s

Throughput & latency

Metric:

#	Model / Maker	Metrics	Parallel	GPU / CPU / RAM	Runtime
1	gemma-3n-E4B-it8BGoogle Performance benchmarkTimebench 3 - Kombi (Prefill + Generation)	277,51 tok/s TG Prefill 4.348 · TTFT 4.623 ms	10×	AMD Radeon AI PRO R9700AMD Ryzen 3 3100 4-Core Processor · 31 GB RAM	llama.cppopenclaw_cliQ8_0	Details →
2	gemma-3n-E4B-it8BGoogle Performance benchmarkTimebench 3 - Kombi (Prefill + Generation)	176,12 tok/s TG Prefill 3.846 · TTFT 2.574 ms	5×	AMD Radeon AI PRO R9700AMD Ryzen 3 3100 4-Core Processor · 31 GB RAM	llama.cppopenclaw_cliQ8_0	Details →
3	gemma-3n-E4B-it8BGoogle Performance benchmarkTimebench 3 - Kombi (Prefill + Generation)	65,22 tok/s TG Prefill 3.462 · TTFT 571 ms	1×	AMD Radeon AI PRO R9700AMD Ryzen 3 3100 4-Core Processor · 31 GB RAM	llama.cppopenclaw_cliQ8_0	Details →

Agent & chat rating

Noch keine Harnessbenchmarks fuer dieses Modell.