RTX 4070
13 records across 3 models (quants: PTQ1_0, Q4_K).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RTX 4070 | Qwen3 14B 14B | Q4_K | llama.cpp | 42.5 | 2099.9 | – | – | 4096 | – | sourced | – | source |
| RTX 4070 | Qwen3 14B 14B | Q4_K | llama.cpp | 32.7 | 1355.8 | – | – | 16384 | – | sourced | – | source |
| RTX 4070 | Qwen3 8B 8B | Q4_K | llama.cpp | 71.2 | 3564.1 | – | – | 4096 | – | sourced | – | source |
| RTX 4070 | Qwen3 8B 8B | Q4_K | llama.cpp | 52.1 | 2064.3 | – | – | 16384 | – | sourced | – | source |
| RTX 4070 | Qwen3 8B 8B | Q4_K | llama.cpp | 38.1 | 1116.7 | – | – | 32768 | – | sourced | – | source |
| RTX 4070 | Ternary-Bonsai-2-27B | PTQ1_0 | llama.cpp (ada-surgery fork) | 102 | 1173 | – | – | 4096 | 2026-10-08 | community | – | source |
| RTX 4070 | Ternary-Bonsai-2-27B | PTQ1_0 | llama.cpp (ada-surgery fork) | 83 | 1137 | – | – | 16384 | 2026-10-08 | community | – | source |
| RTX 4070 | Ternary-Bonsai-2-27B | PTQ1_0 | llama.cpp (ada-surgery fork) | 78 | 1031 | – | – | 32768 | 2026-10-08 | community | – | source |
| RTX 4070 | Ternary-Bonsai-2-27B | PTQ1_0 | llama.cpp (ada-surgery fork) | 69 | 867 | – | – | 65536 | 2026-10-08 | community | – | source |
| RTX 4070 | Ternary-Bonsai-2-27B | PTQ1_0 | llama.cpp (ada-surgery fork) | 61 | 698 | – | – | 114688 | 2026-10-08 | community | – | source |
| RTX 4070 | Ternary-Bonsai-2-27B | PTQ1_0 | llama.cpp (ada-surgery fork) | 30 | 649 | – | – | 131072 | 2026-10-08 | community | – | source |
| RTX 4070 | Ternary-Bonsai-2-27B | PTQ1_0 | llama.cpp (ada-surgery fork) | 17.9 | 547 | – | – | 184320 | 2026-10-08 | community | – | source |
| RTX 4070 | Ternary-Bonsai-2-27B | PTQ1_0 | llama.cpp (ada-surgery fork) | 8.6 | 436 | – | – | 262144 | 2026-10-08 | community | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085