| model                          |       size |     params | backend    | ngl | type_k | type_v |  fa |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -----: | -----: | --: | --------------: | -------------------: |
| minimax-m2 230B.A10B Q4_K - Medium | 131.10 GiB |   228.69 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |           pp512 |     1824.99 ± 597.93 |
| minimax-m2 230B.A10B Q4_K - Medium | 131.10 GiB |   228.69 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |          pp4096 |       3690.55 ± 8.42 |
| minimax-m2 230B.A10B Q4_K - Medium | 131.10 GiB |   228.69 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |           tg128 |        129.47 ± 0.46 |
| minimax-m2 230B.A10B Q4_K - Medium | 131.10 GiB |   228.69 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |  pp512 @ d16384 |       1920.67 ± 6.15 |
| minimax-m2 230B.A10B Q4_K - Medium | 131.10 GiB |   228.69 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 | pp4096 @ d16384 |       2732.74 ± 6.50 |
| minimax-m2 230B.A10B Q4_K - Medium | 131.10 GiB |   228.69 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |  tg128 @ d16384 |         89.70 ± 0.23 |

build: d2462f8f7 (9590)
