| model                          |       size |     params | backend    | ngl | type_k | type_v |  fa |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -----: | -----: | --: | --------------: | -------------------: |
| gpt-oss 120B MXFP4 MoE         |  59.02 GiB |   116.83 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |           pp512 |     9403.28 ± 105.90 |
| gpt-oss 120B MXFP4 MoE         |  59.02 GiB |   116.83 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |          pp4096 |      9096.51 ± 48.52 |
| gpt-oss 120B MXFP4 MoE         |  59.02 GiB |   116.83 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |           tg128 |        275.05 ± 0.83 |
| gpt-oss 120B MXFP4 MoE         |  59.02 GiB |   116.83 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |  pp512 @ d16384 |      7853.99 ± 52.09 |
| gpt-oss 120B MXFP4 MoE         |  59.02 GiB |   116.83 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 | pp4096 @ d16384 |      7495.34 ± 15.72 |
| gpt-oss 120B MXFP4 MoE         |  59.02 GiB |   116.83 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |  tg128 @ d16384 |        235.07 ± 0.44 |

build: d2462f8f7 (9590)