Google Axion, Neoverse V2 with SVE2
6 records across 1 models (quants: Q4_0).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Google Axion, Neoverse V2 with SVE2 | Gemma 4 E4B | Q4_0 | llama.cpp (CPU) | – | 126.69 | – | – | – | 2026-10 | community | – | source |
| Google Axion, Neoverse V2 with SVE2 | Gemma 4 E4B | Q4_0 | llama.cpp (CPU) | – | 91.37 | – | – | – | 2026-10 | community | – | source |
| Google Axion, Neoverse V2 with SVE2 | Gemma 4 E4B | Q4_0 | llama.cpp (CPU) | – | 247.07 | – | – | – | 2026-10 | community | – | source |
| Google Axion, Neoverse V2 with SVE2 | Gemma 4 E4B | Q4_0 | llama.cpp (CPU) | – | 135.1 | – | – | – | 2026-10 | community | – | source |
| Google Axion, Neoverse V2 with SVE2 | Gemma 4 E4B | Q4_0 | llama.cpp (CPU) | – | 121.03 | – | – | – | 2026-10 | community | – | source |
| Google Axion, Neoverse V2 with SVE2 | Gemma 4 E4B | Q4_0 | llama.cpp (CPU) | – | 129.01 | – | – | – | 2026-10 | community | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085