RTX 4070 SUPER
7 records across 3 models (quants: Q4_0, Q4_K).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RTX 4070 SUPER | llama 7B | Q4_0 | llama.cpp (Vulkan) | 112.0 | 5198.91 | – | – | tg128 | 2026-01-16 | community | – | source |
| RTX 4070 SUPER | llama 7B | Q4_0 | llama.cpp (Vulkan) | 108.74 | 4608.2 | – | – | tg128 | 2026-01-16 | community | – | source |
| RTX 4070 SUPER | Qwen3 14B 14B | Q4_K | llama.cpp | 45.5 | 2522.2 | – | – | 4096 | – | sourced | – | source |
| RTX 4070 SUPER | Qwen3 14B 14B | Q4_K | llama.cpp | 37.2 | 1578.1 | – | – | 16384 | – | sourced | – | source |
| RTX 4070 SUPER | Qwen3 8B 8B | Q4_K | llama.cpp | 75.4 | 4321.7 | – | – | 4096 | – | sourced | – | source |
| RTX 4070 SUPER | Qwen3 8B 8B | Q4_K | llama.cpp | 56.2 | 2525.8 | – | – | 16384 | – | sourced | – | source |
| RTX 4070 SUPER | Qwen3 8B 8B | Q4_K | llama.cpp | 42.2 | 1595.7 | – | – | 32768 | – | sourced | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085