| 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 |      4864.86 ± 22.65 |
| qwen3next 80B.A3B Q8_0         |  78.98 GiB |    79.67 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |          pp4096 |       4887.65 ± 6.56 |
| qwen3next 80B.A3B Q8_0         |  78.98 GiB |    79.67 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |           tg128 |        195.06 ± 0.85 |
| qwen3next 80B.A3B Q8_0         |  78.98 GiB |    79.67 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |  pp512 @ d16384 |      4626.57 ± 19.15 |
| qwen3next 80B.A3B Q8_0         |  78.98 GiB |    79.67 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 | pp4096 @ d16384 |       4611.67 ± 5.77 |
| qwen3next 80B.A3B Q8_0         |  78.98 GiB |    79.67 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |  tg128 @ d16384 |        183.70 ± 1.25 |

build: d2462f8f7 (9590)
