deci 70B
12 records across 2 hardware.
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RTX 5060 Ti | deci 70B | IQ4_XS - 4.25 bpw | llama.cpp (CUDA) | 14.6 | 577.31 | – | – | tg128 | 2025-07-30 | community | – | source |
| RTX 5060 Ti | deci 70B | IQ4_XS - 4.25 bpw | llama.cpp (CUDA) | 14.59 | 573.75 | – | – | tg128 | 2025-07-30 | community | – | source |
| RTX 5060 Ti | deci 70B | IQ4_XS - 4.25 bpw | llama.cpp (CUDA) | 14.41 | 549.87 | – | – | tg128 | 2025-07-30 | community | – | source |
| RTX 5060 Ti | deci 70B | IQ4_XS - 4.25 bpw | llama.cpp (CUDA) | 14.35 | 542.01 | – | – | tg128 | 2025-07-30 | community | – | source |
| Arc A770 | deci 70B | IQ4_XS - 4.25 bpw | llama.cpp (ROCm) | 13.33 | 132.43 | – | – | tg128 | 2025-07-30 | community | – | source |
| Arc A770 | deci 70B | IQ4_XS - 4.25 bpw | llama.cpp (ROCm) | 12.89 | 109.97 | – | – | tg128 | 2025-07-30 | community | – | source |
| RTX 5060 Ti | deci 70B | IQ4_XS - 4.25 bpw | llama.cpp (Vulkan) | 10.61 | 112.31 | – | – | tg128 | 2025-07-30 | community | – | source |
| RTX 5060 Ti | deci 70B | IQ4_XS - 4.25 bpw | llama.cpp (Vulkan) | 9.76 | 438.01 | – | – | tg128 | 2025-07-30 | community | – | source |
| RTX 5060 Ti | deci 70B | IQ4_XS - 4.25 bpw | llama.cpp (Vulkan) | 9.73 | 423.66 | – | – | tg128 | 2025-07-30 | community | – | source |
| RTX 5060 Ti | deci 70B | IQ4_XS - 4.25 bpw | llama.cpp (Vulkan) | 8.48 | 426.57 | – | – | tg128 | 2025-07-30 | community | – | source |
| RTX 5060 Ti | deci 70B | IQ4_XS - 4.25 bpw | llama.cpp (Vulkan) | 8.45 | 412.37 | – | – | tg128 | 2025-07-30 | community | – | source |
| RTX 5060 Ti | deci 70B | IQ4_XS - 4.25 bpw | llama.cpp (Vulkan) | 6.73 | 55.4 | – | – | tg128 | 2025-07-30 | community | – | source |