| model                          |       size |     params | backend    | ngl | type_k | type_v |  fa |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -----: | -----: | --: | --------------: | -------------------: |
| qwen3next 80B.A3B Q8_0         |  78.98 GiB |    79.67 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |           pp512 |      4865.77 ± 24.27 |
| qwen3next 80B.A3B Q8_0         |  78.98 GiB |    79.67 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |          pp4096 |       4881.83 ± 9.35 |
| qwen3next 80B.A3B Q8_0         |  78.98 GiB |    79.67 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |           tg128 |        195.02 ± 0.82 |
| qwen3next 80B.A3B Q8_0         |  78.98 GiB |    79.67 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |  pp512 @ d16384 |      4621.06 ± 23.37 |
| qwen3next 80B.A3B Q8_0         |  78.98 GiB |    79.67 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 | pp4096 @ d16384 |       4558.94 ± 9.14 |
| qwen3next 80B.A3B Q8_0         |  78.98 GiB |    79.67 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |  tg128 @ d16384 |        183.62 ± 1.36 |

build: d2462f8f7 (9590)
