Datasets:
LocalLLMAdvisor
/
local-llm-advisor-benchmarks

Dataset card Data Studio Files
xet
 Community
1
hw_id
stringclasses
13 values
device_raw
stringclasses
14 values
device_count
int64
1
1
backend
stringclasses
1 value
model
stringclasses
14 values
quant
stringclasses
4 values
params_b
float64
4.66
117
size_gib
float64
2.54
68
fa
int64
1
1
ngl
int64
99
99
threads
int64
6
128
ubatch
int64
512
512
depth
int64
0
131k
pp_n
int64
130
4.1k
pp_tps
float64
49.3
20.5k
tg_n
int64
64
768
tg_tps
float64
4.18
396
build
stringclasses
1 value
build_num
int64
11.7k
11.7k
n_samples
int64
1
3
date
stringdate
2026-10-16 00:00:00
2026-10-28 00:00:00
notes
stringlengths
250
507
engine
stringclasses
1 value
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Gemma 4 26B-A4B
Q4_K_M
25.81
15.77
1
99
28
512
4,096
512
859.21
128
82.06
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'gemma4 26B.A4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-Q4_K_M.gguf; pp stddev 1...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Gemma 4 26B-A4B
Q4_K_M
25.81
15.77
1
99
28
512
16,384
512
841.41
128
76.38
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'gemma4 26B.A4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-Q4_K_M.gguf; pp stddev 1...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Mistral Small 3.2 24B
Q4_K_M
24.01
13.34
1
99
28
512
0
512
2,306.74
128
52.85
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'llama 13B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Mistral-Small-3.2-24B-Instruct-2506-GGUF/Mistral-Small-3.2-24B-Instruct-250...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Mistral Small 3.2 24B
Q4_K_M
24.01
13.34
1
99
28
512
4,096
512
2,199.84
128
50.3
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'llama 13B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Mistral-Small-3.2-24B-Instruct-2506-GGUF/Mistral-Small-3.2-24B-Instruct-250...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3-Coder 30B-A3B
Q4_K_M
30.53
17.28
1
99
28
512
0
512
691
128
78.15
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen3moe 30B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/Qwen3-Coder-30B-A3B-Instruct-Q4_K_...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3-Coder 30B-A3B
Q4_K_M
30.53
17.28
1
99
28
512
0
512
409.73
128
55.64
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen3moe 30B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/Qwen3-Coder-30B-A3B-Instruct-Q4_K_...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3-Coder 30B-A3B
Q4_K_M
30.53
17.28
1
99
28
512
0
512
297.07
128
39.63
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen3moe 30B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/Qwen3-Coder-30B-A3B-Instruct-Q4_K_...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3-Coder 30B-A3B
Q4_K_M
30.53
17.28
1
99
28
512
4,096
512
703.7
128
81.18
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen3moe 30B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/Qwen3-Coder-30B-A3B-Instruct-Q4_K_...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3-Coder 30B-A3B
Q4_K_M
30.53
17.28
1
99
28
512
16,384
512
643.5
128
73.25
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen3moe 30B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/Qwen3-Coder-30B-A3B-Instruct-Q4_K_...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3-VL 8B
Q4_K_M
8.77
4.68
1
99
28
512
0
512
5,879
128
139.98
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen3vl 8B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-VL-8B-Instruct-GGUF/Qwen3-VL-8B-Instruct-Q4_K_M.gguf; pp stddev 498....
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3-VL 8B
Q4_K_M
8.77
4.68
1
99
28
512
4,096
512
5,126.77
128
124.94
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen3vl 8B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-VL-8B-Instruct-GGUF/Qwen3-VL-8B-Instruct-Q4_K_M.gguf; pp stddev 265....
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3-VL 8B
Q4_K_M
8.77
4.68
1
99
28
512
16,384
512
3,590.3
128
96.19
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen3vl 8B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-VL-8B-Instruct-GGUF/Qwen3-VL-8B-Instruct-Q4_K_M.gguf; pp stddev 135....
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3-VL 8B
Q8_0
8.77
8.11
1
99
28
512
0
512
6,117.64
128
93.3
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen3vl 8B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-VL-8B-Instruct-GGUF/Qwen3-VL-8B-Instruct-Q8_0.gguf; pp stddev 548.31; tg stdd...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3-VL 8B
Q8_0
8.77
8.11
1
99
28
512
4,096
512
5,293.45
128
86.33
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen3vl 8B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-VL-8B-Instruct-GGUF/Qwen3-VL-8B-Instruct-Q8_0.gguf; pp stddev 265.62; tg stdd...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3-VL 8B
Q8_0
8.77
8.11
1
99
28
512
16,384
512
3,697.07
128
72.01
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen3vl 8B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-VL-8B-Instruct-GGUF/Qwen3-VL-8B-Instruct-Q8_0.gguf; pp stddev 140.03; tg stdd...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.5 4B
Q4_K_M
4.66
2.54
1
99
28
512
0
512
8,889.2
128
187.62
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35 4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-4B-GGUF/Qwen3.5-4B-Q4_K_M.gguf; pp stddev 1563.55; HIGH VARIANCE in...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.5 4B
Q4_K_M
4.66
2.54
1
99
28
512
4,096
512
8,265.26
128
179.75
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35 4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-4B-GGUF/Qwen3.5-4B-Q4_K_M.gguf; pp stddev 953.38; HIGH VARIANCE in ...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.5 4B
Q4_K_M
4.66
2.54
1
99
28
512
16,384
512
6,591.77
128
166.78
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35 4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-4B-GGUF/Qwen3.5-4B-Q4_K_M.gguf; pp stddev 627.88; tg stddev 0.99
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.5 4B
Q8_0
4.66
4.16
1
99
28
512
0
512
9,400.32
128
143.41
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35 4B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-4B-GGUF/Qwen3.5-4B-Q8_0.gguf; pp stddev 1426.02; HIGH VARIANCE in pp (>10% o...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.5 4B
Q8_0
4.66
4.16
1
99
28
512
4,096
512
8,767.1
128
139.14
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35 4B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-4B-GGUF/Qwen3.5-4B-Q8_0.gguf; pp stddev 834.21; tg stddev 0.88
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.5 4B
Q8_0
4.66
4.16
1
99
28
512
16,384
512
6,870.43
128
131.4
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35 4B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-4B-GGUF/Qwen3.5-4B-Q8_0.gguf; pp stddev 639.25; tg stddev 0.63
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.5 9B
Q4_K_M
9.65
5.28
1
99
28
512
0
512
5,871.6
128
126.03
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35 9B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q4_K_M.gguf; pp stddev 520.43; tg stddev 0.80
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.5 9B
Q4_K_M
9.65
5.28
1
99
28
512
4,096
512
5,544.27
128
122.67
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35 9B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q4_K_M.gguf; pp stddev 330.41; tg stddev 0.58
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.5 9B
Q4_K_M
9.65
5.28
1
99
28
512
16,384
512
4,729.47
128
116.36
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35 9B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q4_K_M.gguf; pp stddev 290.42; tg stddev 0.50
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.5 9B
Q8_0
9.65
8.86
1
99
28
512
0
512
6,091.48
128
86.39
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35 9B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q8_0.gguf; pp stddev 541.42; tg stddev 0.24
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.5 9B
Q8_0
9.65
8.86
1
99
28
512
4,096
512
5,700.46
128
84.85
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35 9B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q8_0.gguf; pp stddev 453.57; tg stddev 0.28
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.5 9B
Q8_0
9.65
8.86
1
99
28
512
16,384
512
4,837.67
128
81.83
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35 9B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q8_0.gguf; pp stddev 299.01; tg stddev 0.21
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.6 35B-A3B
Q4_K_M
35.95
20.6
1
99
28
512
0
512
405.88
128
68.35
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35moe 35B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q4_K_M.gguf; pp stddev 15.0...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.6 35B-A3B
Q4_K_M
35.95
20.6
1
99
28
512
0
512
289.72
128
53.05
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35moe 35B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q4_K_M.gguf; pp stddev 10.1...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.6 35B-A3B
Q4_K_M
35.95
20.6
1
99
28
512
4,096
512
391.83
128
69.51
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35moe 35B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q4_K_M.gguf; pp stddev 3.60...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
Qwen3.6 35B-A3B
Q4_K_M
35.95
20.6
1
99
28
512
16,384
512
381.88
128
53.73
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'qwen35moe 35B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q4_K_M.gguf; pp stddev 1.78...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
gpt-oss 20B
Q4_K_M
20.91
10.81
1
99
28
512
0
512
9,595.62
128
236.89
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'gpt-oss 20B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q4_K_M.gguf; pp stddev 1354.11; HIGH VARIANC...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
gpt-oss 20B
Q4_K_M
20.91
10.81
1
99
28
512
4,096
512
8,476.64
128
217.64
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'gpt-oss 20B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q4_K_M.gguf; pp stddev 1132.88; HIGH VARIANC...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
gpt-oss 20B
Q4_K_M
20.91
10.81
1
99
28
512
16,384
512
6,736.99
128
208.7
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'gpt-oss 20B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q4_K_M.gguf; pp stddev 435.92; tg stddev 3.5...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
gpt-oss 20B
Q8_0
20.91
11.27
1
99
28
512
0
512
9,839.06
128
223.82
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'gpt-oss 20B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q8_0.gguf; pp stddev 1503.87; HIGH VARIANCE in pp (>1...
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
gpt-oss 20B
Q8_0
20.91
11.27
1
99
28
512
4,096
512
8,778.77
128
207.3
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'gpt-oss 20B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q8_0.gguf; pp stddev 821.31; tg stddev 1.59
llama.cpp
rtx-5070ti
NVIDIA GeForce RTX 5070 Ti
1
CUDA
gpt-oss 20B
Q8_0
20.91
11.27
1
99
28
512
16,384
512
6,858.55
128
198.93
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz, RAM 125 GiB; llama-bench model_type 'gpt-oss 20B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q8_0.gguf; pp stddev 519.00; tg stddev 2.83
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Gemma 4 12B
Q4_K_M
11.96
6.62
1
99
32
512
0
512
4,597.77
128
94.8
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gemma4 ?B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-Q4_K_M.gguf; pp stddev 405.33; tg stddev 0.31
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Gemma 4 12B
Q4_K_M
11.96
6.62
1
99
32
512
4,096
512
4,194.13
128
89.89
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gemma4 ?B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-Q4_K_M.gguf; pp stddev 185.59; tg stddev 0.25
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Gemma 4 12B
Q4_K_M
11.96
6.62
1
99
32
512
16,384
512
3,586.7
128
88.53
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gemma4 ?B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-Q4_K_M.gguf; pp stddev 126.99; tg stddev 0.30
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Gemma 4 12B
Q8_0
11.96
11.78
1
99
32
512
0
512
4,712.82
128
61.81
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gemma4 ?B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-Q8_0.gguf; pp stddev 431.11; tg stddev 0.13
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Gemma 4 12B
Q8_0
11.96
11.78
1
99
32
512
4,096
512
4,312
128
59.65
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gemma4 ?B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-Q8_0.gguf; pp stddev 188.26; tg stddev 0.11
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Gemma 4 12B
Q8_0
11.96
11.78
1
99
32
512
16,384
512
3,673.12
128
59.08
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gemma4 ?B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-Q8_0.gguf; pp stddev 135.15; tg stddev 0.11
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Gemma 4 26B-A4B
Q4_K_M
25.81
15.77
1
99
32
512
0
512
1,304.53
128
65.8
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gemma4 26B.A4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-Q4_K_M.gguf; pp stddev 89.70; t...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Gemma 4 26B-A4B
Q4_K_M
25.81
15.77
1
99
32
512
0
512
797.81
128
43.84
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gemma4 26B.A4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-Q4_K_M.gguf; pp stddev 37.59; t...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Gemma 4 26B-A4B
Q4_K_M
25.81
15.77
1
99
32
512
0
512
647.14
128
32.65
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gemma4 26B.A4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-Q4_K_M.gguf; pp stddev 31.08; t...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Gemma 4 26B-A4B
Q4_K_M
25.81
15.77
1
99
32
512
4,096
512
1,374.06
128
64.04
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gemma4 26B.A4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-Q4_K_M.gguf; pp stddev 19.08; t...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Gemma 4 26B-A4B
Q4_K_M
25.81
15.77
1
99
32
512
16,384
512
1,290.96
128
61.91
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gemma4 26B.A4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-Q4_K_M.gguf; pp stddev 106.29; ...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Mistral Small 3.2 24B
Q4_K_M
24.01
13.34
1
99
32
512
0
512
2,581.12
128
58.64
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'llama 13B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Mistral-Small-3.2-24B-Instruct-2506-GGUF/Mistral-Small-3.2-24B-Instruct-2506-Q4_K_...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Mistral Small 3.2 24B
Q4_K_M
24.01
13.34
1
99
32
512
4,096
512
2,425.44
128
56.04
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'llama 13B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Mistral-Small-3.2-24B-Instruct-2506-GGUF/Mistral-Small-3.2-24B-Instruct-2506-Q4_K_...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3-Coder 30B-A3B
Q4_K_M
30.53
17.28
1
99
32
512
0
512
1,086.09
128
66.93
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen3moe 30B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/Qwen3-Coder-30B-A3B-Instruct-Q4_K_M.gguf;...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3-Coder 30B-A3B
Q4_K_M
30.53
17.28
1
99
32
512
0
512
678.86
128
39.46
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen3moe 30B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/Qwen3-Coder-30B-A3B-Instruct-Q4_K_M.gguf;...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3-Coder 30B-A3B
Q4_K_M
30.53
17.28
1
99
32
512
0
512
492.88
128
28.43
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen3moe 30B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/Qwen3-Coder-30B-A3B-Instruct-Q4_K_M.gguf;...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3-Coder 30B-A3B
Q4_K_M
30.53
17.28
1
99
32
512
4,096
512
1,108.53
128
63.29
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen3moe 30B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/Qwen3-Coder-30B-A3B-Instruct-Q4_K_M.gguf;...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3-Coder 30B-A3B
Q4_K_M
30.53
17.28
1
99
32
512
16,384
512
979.49
128
58.52
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen3moe 30B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/Qwen3-Coder-30B-A3B-Instruct-Q4_K_M.gguf;...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3-VL 8B
Q4_K_M
8.77
4.68
1
99
32
512
0
512
6,837.85
128
149.38
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen3vl 8B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-VL-8B-Instruct-GGUF/Qwen3-VL-8B-Instruct-Q4_K_M.gguf; pp stddev 655.67; tg ...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3-VL 8B
Q4_K_M
8.77
4.68
1
99
32
512
4,096
512
6,014
128
134.92
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen3vl 8B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-VL-8B-Instruct-GGUF/Qwen3-VL-8B-Instruct-Q4_K_M.gguf; pp stddev 298.07; tg ...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3-VL 8B
Q4_K_M
8.77
4.68
1
99
32
512
16,384
512
4,034.12
128
105.49
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen3vl 8B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-VL-8B-Instruct-GGUF/Qwen3-VL-8B-Instruct-Q4_K_M.gguf; pp stddev 130.59; tg ...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3-VL 8B
Q8_0
8.77
8.11
1
99
32
512
0
512
7,048.14
128
98.89
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen3vl 8B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-VL-8B-Instruct-GGUF/Qwen3-VL-8B-Instruct-Q8_0.gguf; pp stddev 806.60; HIGH VARIANCE ...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3-VL 8B
Q8_0
8.77
8.11
1
99
32
512
4,096
512
6,219.92
128
92.42
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen3vl 8B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-VL-8B-Instruct-GGUF/Qwen3-VL-8B-Instruct-Q8_0.gguf; pp stddev 294.50; tg stddev 0.19
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3-VL 8B
Q8_0
8.77
8.11
1
99
32
512
16,384
512
4,143.61
128
77.63
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen3vl 8B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3-VL-8B-Instruct-GGUF/Qwen3-VL-8B-Instruct-Q8_0.gguf; pp stddev 128.86; tg stddev 0.14
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.5 4B
Q4_K_M
4.66
2.54
1
99
32
512
0
512
10,057.3
128
200.46
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35 4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-4B-GGUF/Qwen3.5-4B-Q4_K_M.gguf; pp stddev 1520.80; HIGH VARIANCE in pp (>1...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.5 4B
Q4_K_M
4.66
2.54
1
99
32
512
4,096
512
9,750.07
128
191.66
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35 4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-4B-GGUF/Qwen3.5-4B-Q4_K_M.gguf; pp stddev 650.13; tg stddev 0.93
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.5 4B
Q4_K_M
4.66
2.54
1
99
32
512
16,384
512
8,287.75
128
178.74
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35 4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-4B-GGUF/Qwen3.5-4B-Q4_K_M.gguf; pp stddev 551.23; tg stddev 0.93
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.5 4B
Q8_0
4.66
4.16
1
99
32
512
0
512
10,589.57
128
153.35
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35 4B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-4B-GGUF/Qwen3.5-4B-Q8_0.gguf; pp stddev 1738.31; HIGH VARIANCE in pp (>10% of mean)...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.5 4B
Q8_0
4.66
4.16
1
99
32
512
4,096
512
10,214.8
128
148.04
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35 4B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-4B-GGUF/Qwen3.5-4B-Q8_0.gguf; pp stddev 712.88; tg stddev 0.56
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.5 4B
Q8_0
4.66
4.16
1
99
32
512
16,384
512
8,646.12
128
140.68
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35 4B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-4B-GGUF/Qwen3.5-4B-Q8_0.gguf; pp stddev 604.92; tg stddev 0.58
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.5 9B
Q4_K_M
9.65
5.28
1
99
32
512
0
512
6,543.41
128
136.14
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35 9B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q4_K_M.gguf; pp stddev 711.02; HIGH VARIANCE in pp (>10...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.5 9B
Q4_K_M
9.65
5.28
1
99
32
512
4,096
512
6,403.26
128
131.86
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35 9B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q4_K_M.gguf; pp stddev 342.31; tg stddev 0.50
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.5 9B
Q4_K_M
9.65
5.28
1
99
32
512
16,384
512
5,739.87
128
125.85
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35 9B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q4_K_M.gguf; pp stddev 292.76; tg stddev 0.43
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.5 9B
Q8_0
9.65
8.86
1
99
32
512
0
512
6,790.2
128
92.26
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35 9B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q8_0.gguf; pp stddev 750.41; HIGH VARIANCE in pp (>10% of mean);...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.5 9B
Q8_0
9.65
8.86
1
99
32
512
4,096
512
6,694.94
128
90.33
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35 9B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q8_0.gguf; pp stddev 309.31; tg stddev 0.21
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.5 9B
Q8_0
9.65
8.86
1
99
32
512
16,384
512
5,903.9
128
87.47
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35 9B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q8_0.gguf; pp stddev 365.58; tg stddev 0.20
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.6 35B-A3B
Q4_K_M
35.95
20.6
1
99
32
512
0
512
680.53
128
56.1
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35moe 35B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q4_K_M.gguf; pp stddev 53.20; tg s...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.6 35B-A3B
Q4_K_M
35.95
20.6
1
99
32
512
0
512
501.12
128
41.91
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35moe 35B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q4_K_M.gguf; pp stddev 4.30; tg st...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.6 35B-A3B
Q4_K_M
35.95
20.6
1
99
32
512
4,096
512
688.57
128
56.48
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35moe 35B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q4_K_M.gguf; pp stddev 44.20; tg s...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
Qwen3.6 35B-A3B
Q4_K_M
35.95
20.6
1
99
32
512
16,384
512
647.41
128
52.81
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'qwen35moe 35B.A3B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q4_K_M.gguf; pp stddev 27.30; tg s...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
gpt-oss 20B
Q4_K_M
20.91
10.81
1
99
32
512
0
512
10,825.61
128
253.49
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gpt-oss 20B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q4_K_M.gguf; pp stddev 1577.95; HIGH VARIANCE in pp...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
gpt-oss 20B
Q4_K_M
20.91
10.81
1
99
32
512
4,096
512
10,031.52
128
234.1
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gpt-oss 20B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q4_K_M.gguf; pp stddev 726.36; tg stddev 1.11
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
gpt-oss 20B
Q4_K_M
20.91
10.81
1
99
32
512
16,384
512
8,156.33
128
221.77
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gpt-oss 20B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q4_K_M.gguf; pp stddev 479.80; tg stddev 4.71
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
gpt-oss 20B
Q8_0
20.91
11.27
1
99
32
512
0
512
11,105.78
128
238.83
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gpt-oss 20B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q8_0.gguf; pp stddev 1646.97; HIGH VARIANCE in pp (>10% of m...
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
gpt-oss 20B
Q8_0
20.91
11.27
1
99
32
512
4,096
512
10,297.11
128
221.8
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gpt-oss 20B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q8_0.gguf; pp stddev 754.49; tg stddev 1.20
llama.cpp
rtx-5080
NVIDIA GeForce RTX 5080
1
CUDA
gpt-oss 20B
Q8_0
20.91
11.27
1
99
32
512
16,384
512
8,196.62
128
211.79
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.71.05, vast.ai, CPU AMD EPYC 7D12 32-Core Processor, RAM 125 GiB; llama-bench model_type 'gpt-oss 20B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q8_0.gguf; pp stddev 503.50; tg stddev 2.65
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 12B
Q4_K_M
11.96
6.62
1
99
32
512
0
512
8,523.97
128
152.43
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 ?B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-Q4_K_M.gguf; pp stddev 1285.03; HIGH VARIANCE in p...
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 12B
Q4_K_M
11.96
6.62
1
99
32
512
4,096
512
7,893.63
128
145.24
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 ?B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-Q4_K_M.gguf; pp stddev 611.05; tg stddev 0.59
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 12B
Q4_K_M
11.96
6.62
1
99
32
512
16,384
512
6,818.48
128
143.16
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 ?B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-Q4_K_M.gguf; pp stddev 432.67; tg stddev 0.59
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 12B
Q8_0
11.96
11.78
1
99
32
512
0
512
8,805.63
128
103.67
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 ?B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-Q8_0.gguf; pp stddev 1377.66; HIGH VARIANCE in pp (>10% of ...
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 12B
Q8_0
11.96
11.78
1
99
32
512
4,096
512
8,095.07
128
100.14
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 ?B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-Q8_0.gguf; pp stddev 655.07; tg stddev 0.30
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 12B
Q8_0
11.96
11.78
1
99
32
512
16,384
512
6,921.91
128
99.28
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 ?B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-Q8_0.gguf; pp stddev 455.79; tg stddev 0.28
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 26B-A4B
Q4_K_M
25.81
15.77
1
99
32
512
0
512
11,431.56
128
231.41
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 26B.A4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-Q4_K_M.gguf; pp stddev 2287.16; HI...
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 26B-A4B
Q4_K_M
25.81
15.77
1
99
32
512
4,096
512
10,288.68
128
216.39
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 26B.A4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-Q4_K_M.gguf; pp stddev 1329.66; HI...
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 26B-A4B
Q4_K_M
25.81
15.77
1
99
32
512
16,384
512
9,261.99
128
213.02
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 26B.A4B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-Q4_K_M.gguf; pp stddev 851.28; tg ...
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 26B-A4B
Q8_0
25.81
25
1
99
32
512
0
512
11,322.19
128
217.48
3e4dfa8f3
11,658
3
2026-10-18
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 26B.A4B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-Q8_0.gguf; pp stddev 2357.32; HIGH VARIANCE in...
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 26B-A4B
Q8_0
25.81
25
1
99
32
512
4,096
512
10,335.09
128
204.47
3e4dfa8f3
11,658
3
2026-10-18
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 26B.A4B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-Q8_0.gguf; pp stddev 1170.46; HIGH VARIANCE in...
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 26B-A4B
Q8_0
25.81
25
1
99
32
512
16,384
512
9,257.66
128
201.66
3e4dfa8f3
11,658
3
2026-10-18
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 26B.A4B Q8_0'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-Q8_0.gguf; pp stddev 780.33; tg stddev 1.60
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 31B
Q4_K_M
31.27
17.05
1
99
32
512
0
512
3,841.2
128
71.3
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 31B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-31B-it-GGUF/gemma-4-31B-it-Q4_K_M.gguf; pp stddev 293.07; tg stddev 0.23
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 31B
Q4_K_M
31.27
17.05
1
99
32
512
4,096
512
3,606.52
128
67.18
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 31B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-31B-it-GGUF/gemma-4-31B-it-Q4_K_M.gguf; pp stddev 150.53; tg stddev 0.18
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Gemma 4 31B
Q4_K_M
31.27
17.05
1
99
32
512
16,384
512
2,998.41
128
64.61
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'gemma4 31B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/gemma-4-31B-it-GGUF/gemma-4-31B-it-Q4_K_M.gguf; pp stddev 102.48; tg stddev 0.18
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Mistral Small 3.2 24B
Q4_K_M
24.01
13.34
1
99
32
512
0
512
5,184.84
128
101.14
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'llama 13B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Mistral-Small-3.2-24B-Instruct-2506-GGUF/Mistral-Small-3.2-24B-Instruct-2506-Q4_K_M.g...
llama.cpp
rtx-5090
NVIDIA GeForce RTX 5090
1
CUDA
Mistral Small 3.2 24B
Q4_K_M
24.01
13.34
1
99
32
512
4,096
512
5,053.08
128
96.2
3e4dfa8f3
11,658
3
2026-10-17
own run, llama-bench from llama.cpp b11658 (commit 3e4dfa8f3), CUDA 12.8 build, driver 595.84, vast.ai, CPU AMD EPYC 7532 32-Core Processor, RAM 188 GiB; llama-bench model_type 'llama 13B Q4_K - Medium'; -r 3; GPU 0 only; GGUF unsloth/Mistral-Small-3.2-24B-Instruct-2506-GGUF/Mistral-Small-3.2-24B-Instruct-2506-Q4_K_M.g...
llama.cpp