Dataset Viewer
Auto-converted to Parquet Duplicate
model_slug
stringlengths
8
45
display_name
stringlengths
5
37
family
stringclasses
21 values
org
stringclasses
18 values
params_b
float64
0.14
117
⌀
active_params_b
float64
1
17
⌀
arch
stringclasses
2 values
license
stringclasses
16 values
quantization
stringclasses
5 values
ollama_ref
stringlengths
11
52
model_version_id
int64
1
239
run_id
stringclasses
12 values
battery_version
stringclasses
1 value
tested_at
stringclasses
12 values
ranked
bool
2 classes
overall
float64
15.3
99.3
⌀
overall_ci_low
float64
11.7
98.3
⌀
overall_ci_high
float64
19.1
100
⌀
hard_score
float64
6.39
94.7
⌀
hard_n
float64
0
71
⌀
overall_with_second_chance
float64
15.9
99.6
⌀
n_truncated
int64
0
119
n_extended
int64
0
106
n_gated
int64
0
16
fell_for_gotchas
int64
0
14
judge_a_model
stringclasses
2 values
judge_a_mean
float64
1.3
8.78
judge_b_model
stringclasses
2 values
judge_b_mean
float64
0.69
8.4
judge_delta_mean
float64
0.44
1.42
judge_agreement
stringclasses
1 value
n_disputed
int64
0
25
backend
stringclasses
7 values
gpu
stringclasses
1 value
concurrency
stringclasses
5 values
context_tokens
int64
2.05k
16.4k
category_reasoning
float64
2.86
97.1
category_math
float64
0
100
category_coding
float64
0
99.9
category_instruction
float64
13.9
100
category_knowledge
float64
3.33
100
category_long_context
float64
0
100
category_multilingual
float64
10.8
100
category_tool_use
float64
10
100
category_safety
float64
16.7
100
category_sycophancy
float64
25
100
category_gotchas
float64
6.67
100
category_writing
float64
category_identity
float64
speed_decode_tps
float64
1.37
275
⌀
speed_decode_tps_c4
float64
2.65
261
speed_ttft_s
float64
0.02
4.3
⌀
speed_prefill_tps_4k
float64
15.5
9.53k
⌀
speed_prefill_tps_12k
float64
11.4
6.5k
⌀
speed_load_seconds
float64
3.04
463
⌀
speed_size_gb
float64
0.59
76.8
⌀
speed_vram_gb
float64
0.59
76.8
⌀
speed_gpu_fraction
float64
0.9
1
⌀
speed_context_length
float64
2.05k
16.4k
⌀
google-gemma-4-31b-it
gemma-4-31B-it
gemma
google
31.27
null
dense
apache-2.0
Q4_K_M
gemma4:31b-it-q4_K_M
12
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
99.3091
98.33
99.99
94.7151
38
99.6266
3
3
0
1
openai/gpt-5.6-luna
8.779
anthropic/claude-haiku-5.5
8.2964
0.9302
high
2
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
4 at a time
16,384
97.1429
100
96.6392
100
100
100
100
100
100
100
93.3333
null
null
3.8389
11.7935
2.3682
63.6962
57.5627
122.6714
11.1594
11.1594
1
16,384
openai-gpt-oss-120b
gpt-oss-120b
gpt-oss
openai
116.83
5.1
moe
apache-2.0
MXFP4
gpt-oss:120b
67
r20261012-bv1-6a1291
v1
2026-10-14T22:08:38.409Z
true
98.606
96.78
99.99
88.6667
18
98.606
1
1
0
0
anthropic/claude-haiku-5.5
7.911
deepseek/deepseek-v4.1-flash
8.402
0.7112
high
2
Vulkan (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
4 at a time
16,384
97.1429
96
99.8667
100
100
94.4444
100
100
100
91.6667
100
null
null
11.0328
15.8198
2.7321
157.7708
140.1032
104.759
64.6046
64.6046
1
16,384
google-gemma-4-12b-it
Gemma 4 12B IT
gemma
google
12
null
dense
apache-2.0
Q4_K_M
gemma4:12b-it-q4_K_M
10
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
98.5523
96.98
99.79
90.7165
38
99.1696
4
4
0
1
openai/gpt-5.6-luna
8.6748
anthropic/claude-haiku-5.5
8.0714
0.9167
high
4
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
94.2857
100
98.2409
100
100
94.4444
100
100
100
100
93.3333
null
null
8.119
13.1275
0.9368
156.8684
147.8502
89.0233
8.9078
8.9078
1
16,384
muse-glimmer-30b
Muse Glimmer 30B
other
null
30
null
dense
null
Q4_K_M
muse-glimmer:30b-q4_K_M
222
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
97.8982
96.03
99.32
84.8744
35
99.2959
10
10
0
0
openai/gpt-5.6-luna
8.639
anthropic/claude-haiku-5.5
8.1821
0.8612
high
3
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
94.2857
100
92.3534
100
100
94.4444
100
100
100
100
100
null
null
3.6388
8.8104
2.1202
96.0664
96.4292
42.2923
18.329
18.329
1
16,384
google-gemma-4-26b-a4b-it
gemma-4-26B-A4B-it
gemma
google
26
4
moe
apache-2.0
Q4_K_M
gemma4:26b-a4b-it-q4_K_M
11
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
97.7523
95.76
99.32
92.0494
38
97.7523
5
5
0
0
openai/gpt-5.6-luna
8.5362
anthropic/claude-haiku-5.5
8.2717
0.7498
high
0
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
94.2857
100
96.5959
100
100
88.8889
100
100
100
100
100
null
null
15.7103
21.8342
0.615
365.6299
330.4931
70.0669
22.6369
22.6369
1
16,384
qwen-qwen3-coder-next
Qwen3-Coder-Next
qwen
Qwen
79.67
3
moe
apache-2.0
Q4_K_M
qwen3-coder-next:q4_K_M
68
r20261012-bv1-6a1291
v1
2026-10-14T22:08:38.409Z
true
96.3845
94.18
98.25
66.5739
18
98.2257
16
16
0
0
openai/gpt-5.6-luna
8.3074
anthropic/claude-haiku-5.5
7.7714
0.8364
high
5
Vulkan (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
88.5714
92
97.2777
95.1667
100
94.4444
100
100
100
91.6667
100
null
null
18.1009
14.6929
1.9263
177.2516
179.5863
75.7438
52.2472
52.2472
1
16,384
qwen-qwen3-5-27b
Qwen3.5-27B
qwen
Qwen
27.78
null
dense
apache-2.0
Q4_K_M
qwen3.5:27b-q4_K_M
50
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
96.1888
93.82
98.26
78.7851
38
98.4745
20
20
0
1
openai/gpt-5.6-luna
8.4362
anthropic/claude-haiku-5.5
7.9879
0.9064
high
2
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
88.5714
84
93.1278
100
100
100
100
100
100
100
93.3333
null
null
2.7213
2.648
2.0584
70.5111
69.0165
27.6661
17.6015
17.6015
1
16,384
google-gemma-3-27b-it
gemma-3-27b-it
gemma
google
27.43
null
dense
gemma
Q4_K_M
gemma3:27b-it-q4_K_M
7
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
95.7673
93.7
97.53
76.0127
38
95.6218
23
21
0
1
openai/gpt-5.6-luna
7.7626
anthropic/claude-haiku-5.5
7.3
1.114
high
11
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
91.4286
96
85.3938
94.5
96.6667
100
100
100
97.9167
100
93.3333
null
null
2.7226
4.6898
1.9135
70.8791
71.5421
124.9378
22.0768
22.0768
1
16,384
qwen-qwen3-coder-30b-a3b-instruct
Qwen3-Coder-30B-A3B-Instruct
qwen
Qwen
30
3
moe
apache-2.0
Q4_K_M
qwen3-coder:30b-a3b-q4_K_M
58
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
95.5399
93.18
97.56
72.9025
38
95.5399
3
3
0
3
openai/gpt-5.6-luna
7.8164
anthropic/claude-haiku-5.5
7.4829
0.9045
high
8
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
88.5714
92
91.9543
92.8889
100
94.4444
100
100
100
83.3333
80
null
null
20.0553
26.0777
0.4889
382.3662
302.973
100.4754
31.6801
31.6801
1
16,384
gemma4-e4b-it
Gemma4 E4B IT
gemma
null
8
4
dense
apache-2.0
Q4_K_M
gemma4:e4b-it-q4_K_M
106
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
95.4613
92.97
97.72
84.7152
71
97.5071
16
16
1
1
openai/gpt-5.6-luna
7.8817
anthropic/claude-haiku-5.5
7.6726
0.7933
high
2
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
85.7143
96
89.937
97.2222
100
94.4444
100
100
95.8333
91.6667
93.3333
null
null
13.1584
25.3133
0.4079
416.4618
390.672
30.469
1.1893
1.1893
1
16,384
qwen-qwen3-8-27b
Qwen3.8-27B
qwen
Qwen
27.78
null
dense
apache-2.0
Q4_K_M
qwen3.8:27b-q4_K_M
56
r20261012-bv1-6a1291
v1
2026-10-14T22:08:38.409Z
true
95.4183
92.87
97.75
72.5443
18
98.3213
21
21
0
1
openai/gpt-5.6-luna
8.1205
anthropic/claude-haiku-5.5
7.9226
0.7955
high
6
Vulkan (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
88.5714
80
96.8599
98.8889
100
94.4444
100
100
100
100
93.3333
null
null
3.4517
3.1692
4.1338
51.919
51.2641
12.2341
17.6224
17.6224
1
16,384
qwen-qwen3-6-27b
Qwen3.6-27B
qwen
Qwen
27.78
null
dense
apache-2.0
Q4_K_M
qwen3.6:27b-q4_K_M
54
r20261012-bv1-6a1291
v1
2026-10-14T22:08:38.409Z
true
95.0871
92.4
97.51
57.1042
18
98.2158
19
19
0
1
openai/gpt-5.6-luna
8.0924
anthropic/claude-haiku-5.5
7.7452
0.78
high
2
Vulkan (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
88.5714
88
91.9898
98.3333
100
88.8889
100
100
100
100
93.3333
null
null
null
3.4518
null
55.246
54.5776
11.8451
17.3089
17.3089
1
16,384
qwen-qwen3-30b-a3b-instruct-2507
Qwen3-30B-A3B-Instruct-2507
qwen
Qwen
30
3
moe
apache-2.0
Q4_K_M
qwen3:30b-a3b-instruct-2507-q4_K_M
46
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
95.0349
92.51
97.24
73.1738
38
96.4317
14
14
0
2
openai/gpt-5.6-luna
7.9926
anthropic/claude-haiku-5.5
7.4836
0.9781
high
7
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
88.5714
96
93.1868
90.3333
100
88.8889
98.3333
100
100
100
86.6667
null
null
20.36
26.1234
0.4904
383.5951
303.9226
100.2712
31.6801
31.6801
1
16,384
openai-gpt-oss-20b
gpt-oss-20b
gpt-oss
openai
20.91
3.6
moe
apache-2.0
MXFP4
gpt-oss:20b
237
r20261105-bv1-2abe78
v1
2026-11-14T12:26:35.367Z
true
95.026
92.15
97.57
null
0
96.9704
11
11
0
0
anthropic/claude-haiku-5.5
6.7002
deepseek/deepseek-v4.1-flash
7.0367
0.6255
high
0
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
97.1429
96
95.1467
95.8333
100
94.4444
100
97.5
79.1667
100
100
null
null
14.8812
31.4864
0.6209
437.1095
429.0446
40.5584
13.5285
13.5285
1
16,384
qwen-qwen3-vl-8b-instruct
Qwen3-VL-8B-Instruct
qwen
Qwen
8.77
null
dense
apache-2.0
Q4_K_M
qwen3-vl:8b-instruct-q4_K_M
62
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
94.8954
92.4
97.09
66.0675
38
95.7433
13
13
0
1
openai/gpt-5.6-luna
7.5848
anthropic/claude-haiku-5.5
7.2476
0.9767
high
4
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
91.4286
92
77.6855
98.5
100
94.4444
100
100
100
100
93.3333
null
null
9.3756
9.1811
0.4322
261.4355
228.9543
25.868
7.7609
7.7609
1
16,384
qwen-qwen3-5-35b-a3b
Qwen3.5-35B-A3B
qwen
Qwen
35
3
moe
apache-2.0
Q4_K_M
qwen3.5:35b-a3b-q4_K_M
51
r20261012-bv1-6a1291
v1
2026-10-14T22:08:38.409Z
true
94.8139
92.1
97.08
41.8012
18
96.8404
18
18
0
1
openai/gpt-5.6-luna
8.1405
anthropic/claude-haiku-5.5
7.7057
0.8495
high
7
Vulkan (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
85.7143
92
87.111
93.5
100
100
100
95
100
100
93.3333
null
null
23.1804
16.4638
1.4125
235.0194
226.924
15.2613
21.7179
21.7179
1
16,384
qwen-qwen3-vl-32b-instruct
Qwen3-VL-32B-Instruct
qwen
Qwen
33.36
null
dense
apache-2.0
Q4_K_M
qwen3-vl:32b-instruct-q4_K_M
60
r20261012-bv1-6a1291
v1
2026-10-14T22:08:38.409Z
true
94.6605
92.09
96.92
51.035
18
96.5018
12
12
0
1
openai/gpt-5.6-luna
7.8867
anthropic/claude-haiku-5.5
7.5598
0.9431
high
8
Vulkan (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
88.5714
88
86.5402
96.0556
100
94.4444
98.3333
100
100
100
93.3333
null
null
3.1632
2.8913
1.4851
32.7342
21.6184
15.6193
24.5243
24.5243
1
16,384
qwen-qwen3-6-35b-a3b
Qwen3.6-35B-A3B
qwen
Qwen
35
3
moe
apache-2.0
Q4_K_M
qwen3.6:35b-a3b-q4_K_M
55
r20261012-bv1-6a1291
v1
2026-10-14T22:08:38.409Z
true
94.6282
91.57
97.21
58.4173
18
97.779
21
21
0
1
openai/gpt-5.6-luna
8.1981
anthropic/claude-haiku-5.5
7.7129
0.9748
high
4
Vulkan (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
85.7143
92
94.0504
96.5556
100
83.3333
100
100
100
91.6667
93.3333
null
null
22.0703
17.8359
1.6289
276.3907
284.0666
15.1675
23.0478
23.0478
1
16,384
internscience-sage-a1-5-q4-k-m-latest
Sage-A1.5-Q4_K_M
other
null
36
null
null
null
Q4_K_M
hf.co/InternScience/Sage-A1.5-Q4_K_M-GGUF:latest
212
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
94.4719
91.83
96.74
64.103
35
97.3749
17
17
0
1
openai/gpt-5.6-luna
8.0798
anthropic/claude-haiku-5.5
7.5788
1.0419
high
9
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
88.5714
88
84.6201
94.6111
100
94.4444
100
100
100
100
93.3333
null
null
22.5502
21.2185
0.5693
395.0861
374.9531
25.1765
21.783
21.783
1
16,384
meta-llama-llama-3-3-70b-instruct
Llama-3.3-70B-Instruct
llama
meta-llama
70.55
null
dense
llama3.3
Q4_K_M
llama3.3:70b-instruct-q4_K_M
65
r20261012-bv1-6a1291
v1
2026-10-14T22:08:38.409Z
true
94.2509
91.31
96.71
62.574
18
94.2509
1
1
0
1
openai/gpt-5.6-luna
8.119
anthropic/claude-haiku-5.5
7.515
0.9669
high
5
Vulkan (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
4 at a time
16,384
97.1429
84
80.7262
97.5
100
88.8889
100
100
100
100
93.3333
null
null
1.3738
3.4367
4.3013
15.5233
11.4146
51.0511
63.8494
63.8494
1
16,384
granite4-small-h
Granite4 Small H
granite
null
32
9
moe
null
Q4_K_M
granite4:small-h
211
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
93.9127
91.11
96.49
66.4555
35
93.9127
0
0
0
3
openai/gpt-5.6-luna
7.8305
anthropic/claude-haiku-5.5
7.346
0.9845
high
9
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
88.5714
88
86.3092
93.4444
100
88.8889
100
100
100
100
80
null
null
8.4433
12.3916
1.4353
133.1758
135.4183
37.9266
23.9189
23.9189
1
16,384
gemma4-e2b-it
Gemma4 E2B IT
gemma
null
5
2
dense
apache-2.0
Q4_K_M
gemma4:e2b-it-q4_K_M
105
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
93.7501
91.03
96.12
79.7941
71
94.9749
15
15
0
1
openai/gpt-5.6-luna
7.834
anthropic/claude-haiku-5.5
7.3657
1.0321
high
5
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
88.5714
88
80.0682
95.4444
96.6667
100
100
95
100
91.6667
93.3333
null
null
17.153
29.816
0.3947
672.8474
593.0239
18.344
0.5929
0.5929
1
16,384
qwen-qwen3-32b
Qwen3-32B
qwen
Qwen
32.76
null
dense
apache-2.0
Q4_K_M
qwen3:32b-q4_K_M
47
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
93.6304
90.91
96.3
64.7483
38
94.9002
9
9
0
2
openai/gpt-5.6-luna
7.8919
anthropic/claude-haiku-5.5
7.4986
0.88
high
6
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
85.7143
88
83.7924
96.2778
100
88.8889
100
100
100
100
86.6667
null
null
2.2793
4.1371
1.7561
54.297
45.5854
246.2199
54.9647
54.9647
1
16,384
meta-llama-llama-4-scout-17b-16e-instruct
Llama 4 Scout 17B-16E Instruct
llama
meta-llama
109
17
moe
llama4
Q4_K_M
llama4:17b-scout-16e-instruct-q4_K_M
66
r20261012-bv1-6a1291
v1
2026-10-14T22:08:38.409Z
true
93.5581
90.73
96.12
69.9401
18
93.5581
0
0
0
1
openai/gpt-5.6-luna
7.7229
anthropic/claude-haiku-5.5
7.4376
0.9614
high
5
Vulkan (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
91.4286
84
80.9834
98.6667
100
94.4444
100
92.5
100
100
93.3333
null
null
16.9535
8.2324
2.5258
147.436
162.8656
81.4357
70.3669
70.3669
1
16,384
qwen-qwen2-5-32b-instruct
Qwen2.5-32B-Instruct
qwen
Qwen
32.76
null
dense
apache-2.0
Q4_K_M
qwen2.5:32b-instruct-q4_K_M
33
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
93.5071
90.76
96.13
61.8109
38
93.5071
0
0
0
3
openai/gpt-5.6-luna
7.949
anthropic/claude-haiku-5.5
7.3667
1.0286
high
6
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
85.7143
92
81.9605
94.1111
100
94.4444
93.3333
100
100
100
80
null
null
2.3043
7.6125
1.6149
52.5156
46.1117
251.6695
54.2034
54.2034
1
16,384
qwen-qwen3-vl-30b-a3b-instruct
Qwen3-VL-30B-A3B-Instruct
qwen
Qwen
30
3
moe
apache-2.0
Q4_K_M
qwen3-vl:30b-a3b-instruct-q4_K_M
59
r20261012-bv1-6a1291
v1
2026-10-14T22:08:38.409Z
true
92.8857
90.06
95.53
43.8632
18
95.6617
16
16
0
3
openai/gpt-5.6-luna
7.776
anthropic/claude-haiku-5.5
7.3729
0.8579
high
6
Vulkan (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
82.8571
88
87.8365
88.3889
100
88.8889
100
100
100
83.3333
80
null
null
26.9071
19.8201
0.7368
297.6441
166.8251
12.597
20.6197
20.6197
1
16,384
mistralai-mistral-small-3-2-24b-instruct-2506
Mistral Small 3.2 24B Instruct (2506)
mistral
mistralai
24
null
dense
apache-2.0
Q4_K_M
mistral-small3.2:24b-instruct-2506-q4_K_M
28
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
92.8816
90.25
95.15
61.7279
38
94.2785
4
4
0
1
openai/gpt-5.6-luna
7.7683
anthropic/claude-haiku-5.5
7.1088
1.079
high
7
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
82.8571
96
77.1887
85.4444
100
94.4444
100
100
100
100
93.3333
null
null
3.1182
9.3526
1.7174
98.934
79.6846
176.9068
36.4783
36.4783
1
16,384
nemotron-3-5-lightning-30b-a3b
Nemotron 3.5 Lightning 30B A3B
nemotron
null
30
3
moe
null
Q4_K_M
nemotron-3.5-lightning:30b-a3b-q4_K_M
224
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
92.8242
89.69
95.76
65.6507
35
93.5861
9
9
0
0
openai/gpt-5.6-luna
7.7555
anthropic/claude-haiku-5.5
7.324
1.0448
high
10
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
85.7143
92
89.0369
93.6667
100
83.3333
96.6667
95
100
91.6667
100
null
null
17.7208
17.3384
0.8495
352.192
351.5594
22.6022
25.4364
25.4364
1
16,384
granite4-2-30b
Granite4.2 30B
granite
null
30
null
dense
null
Q4_K_M
granite4.2:30b-q4_K_M
208
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
92.8069
89.89
95.41
67.0953
35
96.9162
27
26
0
1
openai/gpt-5.6-luna
7.6755
anthropic/claude-haiku-5.5
7.4438
1.0164
high
13
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
80
80
87.9842
92.8333
100
94.4444
100
100
100
91.6667
93.3333
null
null
3.3926
8.2145
1.3651
62.0325
59.6169
243.2704
52.3497
52.3497
1
16,384
zai-org-glm-4-5-air
GLM-4.5-Air
glm
zai-org
110.47
12
moe
mit
Q4_K_M
hf.co/unsloth/GLM-4.5-Air-GGUF:Q4_K_M
69
r20261015-bv1-5ac997
v1
2026-10-15T13:45:35.381Z
true
92.7568
89.66
95.61
null
0
94.3441
27
24
0
1
openai/gpt-5.6-luna
7.4133
anthropic/claude-haiku-5.5
7.1664
1.0569
high
8
Vulkan (Ollama 0.40.4, flash attention, 16k context; 2 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
2 at a time
16,384
82.8571
96
87.4541
94.3333
100
83.3333
95
100
95.8333
75
93.3333
null
null
5.9864
7.5927
3.1499
53.4418
28.8813
110.5245
76.7538
76.7538
1
16,384
qwen-qwen3-30b-a3b
Qwen3-30B-A3B
qwen
Qwen
30
3
moe
apache-2.0
Q4_K_M
qwen3:30b-a3b-q4_K_M
45
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
92.7502
89.38
95.57
65.816
38
93.3851
6
6
0
1
openai/gpt-5.6-luna
7.784
anthropic/claude-haiku-5.5
7.2643
0.9245
high
8
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
88.5714
84
86.5689
92.2778
100
83.3333
100
100
100
91.6667
93.3333
null
null
19.6786
25.4957
0.5183
379.7907
301.0756
101.3695
31.7459
31.7459
1
16,384
google-gemma-3-12b-it
gemma-3-12b-it
gemma
google
12.19
null
dense
gemma
Q4_K_M
gemma3:12b-it-q4_K_M
6
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
92.742
89.65
95.37
66.5313
38
93.0594
23
20
0
5
openai/gpt-5.6-luna
7.431
anthropic/claude-haiku-5.5
7.0826
0.9864
high
12
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
88.5714
96
81.8285
88.2778
96.6667
83.3333
100
100
100
91.6667
66.6667
null
null
6.1155
9.4774
0.843
139.0193
138.4609
115.3567
13.0607
13.0607
1
16,384
mistralai-ministral-3-14b-instruct-2512
Ministral 3 14B Instruct (2512)
mistral
mistralai
14
null
dense
apache-2.0
Q4_K_M
ministral-3:14b-instruct-2512-q4_K_M
26
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
92.7343
89.97
95.17
62.3918
38
94.0042
16
16
0
2
openai/gpt-5.6-luna
7.4357
anthropic/claude-haiku-5.5
6.9957
1.0814
high
14
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
82.8571
88
87.5296
81.2222
96.6667
100
98.3333
100
100
100
86.6667
null
null
5.6275
15.0083
0.7308
177.7442
134.7123
173.0056
29.9808
29.9808
1
16,384
qwen-qwq-32b
QwQ 32B
qwen
Qwen
32.8
null
dense
apache-2.0
Q4_K_M
qwq:32b-q4_K_M
239
r20261105-bv1-2abe78
v1
2026-11-14T12:26:35.367Z
true
92.6424
89.82
95.35
null
0
97.6784
17
14
0
0
openai/gpt-5.6-luna
7.4152
anthropic/claude-haiku-5.5
7.0067
0.8376
high
7
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
94.2857
88
60.6074
95.0556
100
100
100
100
95.8333
100
100
null
null
3.1549
4.7182
1.4579
72.092
60.1915
271.7463
54.2244
54.2244
1
16,384
granite4-1-30b
Granite4.1 30B
granite
null
30
null
dense
null
Q4_K_M
granite4.1:30b-q4_K_M
207
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
92.3162
89.29
95.26
62.1804
35
92.6336
3
3
0
2
openai/gpt-5.6-luna
7.8376
anthropic/claude-haiku-5.5
7.4383
0.8607
high
4
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
82.8571
96
85.9883
93.7778
100
72.2222
100
100
100
91.6667
86.6667
null
null
3.3861
5.0011
1.4861
62.4365
59.1875
242.0584
52.3497
52.3497
1
16,384
granite4-1-8b
Granite4.1 8B
granite
null
8
null
dense
apache-2.0
Q4_K_M
granite4.1:8b-q4_K_M
122
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
92.1412
88.96
95
76.7792
71
92.1412
0
0
0
3
openai/gpt-5.6-luna
7.6017
anthropic/claude-haiku-5.5
7.3221
0.9405
high
7
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
91.4286
88
76.4532
93.1111
96.6667
94.4444
93.3333
100
95.8333
91.6667
80
null
null
10.436
15.9417
0.5004
224.8955
183.7273
189.687
27.1039
27.1039
1
16,384
mistral-small-24b-instruct-2501
Mistral Small 24B Instruct 2501
mistral
null
24
null
dense
null
Q4_K_M
mistral-small:24b-instruct-2501-q4_K_M
219
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
92.138
89.3
94.69
53.8471
35
92.138
0
0
0
2
openai/gpt-5.6-luna
7.6371
anthropic/claude-haiku-5.5
7.066
0.8607
high
7
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
85.7143
84
82.7502
84.8333
100
94.4444
97.5
100
100
100
86.6667
null
null
4.2181
10.5787
1.032
98.4382
87.8892
204.3266
36.4636
36.4636
1
16,384
mistralai-devstral-small-2-24b-instruct-2512
Devstral Small 2 24B Instruct (2512)
mistral
mistralai
24
null
dense
apache-2.0
Q4_K_M
devstral-small-2:24b-instruct-2512-q4_K_M
25
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
92.0736
89.31
94.71
63.1606
38
93.2799
9
9
0
3
openai/gpt-5.6-luna
7.8129
anthropic/claude-haiku-5.5
7.2126
1.0198
high
9
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
85.7143
76
83.3924
89.1111
100
94.4444
100
100
100
100
80
null
null
3.1363
8.8856
1.5395
79.9954
64.8535
177.645
36.4783
36.4783
1
16,384
qwen-qwen3-5-9b
Qwen3.5-9B
qwen
Qwen
9.65
null
dense
apache-2.0
Q4_K_M
qwen3.5:9b-q4_K_M
53
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
92.039
88.99
95.12
59.8885
38
95.2418
17
17
0
1
openai/gpt-5.6-luna
7.7495
anthropic/claude-haiku-5.5
7.4607
1.1369
high
13
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
85.7143
84
77.3588
94.0556
100
88.8889
98.3333
100
100
100
93.3333
null
null
8.4176
8.2368
0.6868
225.4248
221.6891
13.5315
6.1139
6.1139
1
16,384
qwen-qwen3-14b
Qwen3-14B
qwen
Qwen
14.77
null
dense
apache-2.0
Q4_K_M
qwen3:14b-q4_K_M
44
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
91.707
88.48
94.73
52.9304
38
92.3419
10
10
0
3
openai/gpt-5.6-luna
7.7502
anthropic/claude-haiku-5.5
7.3095
0.9083
high
6
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
91.4286
88
73.4897
93.5556
100
88.8889
90
100
100
91.6667
80
null
null
4.9961
8.6657
0.8052
153.4525
112.9015
171.4608
30.9283
30.9283
1
16,384
qwen-qwen3-8b
Qwen3-8B
qwen
Qwen
8.19
null
dense
apache-2.0
Q4_K_M
qwen3:8b-q4_K_M
57
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
91.5462
88.37
94.5
55.4539
38
91.8637
9
6
0
2
openai/gpt-5.6-luna
7.3138
anthropic/claude-haiku-5.5
6.7726
0.9883
high
9
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
94.2857
88
70.5749
87.7222
100
83.3333
100
100
100
100
86.6667
null
null
9.2875
14.3728
0.4989
249.4424
202.9387
148.5704
24.6665
24.6665
1
16,384
qwen-qwen2-5-14b-instruct
Qwen2.5-14B-Instruct
qwen
Qwen
14.77
null
dense
apache-2.0
Q4_K_M
qwen2.5:14b-instruct-q4_K_M
32
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
91.2781
88.01
94.22
63.63
38
91.2781
0
0
0
2
openai/gpt-5.6-luna
7.6212
anthropic/claude-haiku-5.5
7.1121
1.0214
high
14
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
88.5714
84
76.4869
91.3333
100
94.4444
91.6667
95
100
91.6667
86.6667
null
null
21.67
59.0027
0.1831
555.2244
415.7516
141.0471
34.8467
34.8467
1
16,384
laguna-xs-2-1
Laguna XS 2.1
other
null
null
null
null
null
Q4_K_M
laguna-xs-2.1:q4_K_M
213
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
91.1849
87.65
94.37
74.3924
35
91.1849
1
1
0
1
openai/gpt-5.6-luna
7.9569
anthropic/claude-haiku-5.5
7.6205
0.9412
high
8
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
91.4286
88
88.5134
93.2778
96.6667
94.4444
98.3333
70
100
91.6667
93.3333
null
null
23.1114
40.0182
0.642
323.363
288.9103
68.3955
23.1855
23.1855
1
16,384
qwen-qwen2-5-coder-32b-instruct
Qwen2.5-Coder-32B-Instruct
qwen
Qwen
32.76
null
dense
apache-2.0
Q4_K_M
qwen2.5-coder:32b-instruct-q4_K_M
37
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
90.5357
87.25
93.76
64.9031
38
90.5357
2
2
0
3
openai/gpt-5.6-luna
7.6569
anthropic/claude-haiku-5.5
7.355
0.8829
high
7
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
80
64
86.5439
94.8333
100
94.4444
95
100
100
91.6667
80
null
null
2.3009
7.6424
1.6116
51.762
45.8706
244.5076
54.2034
54.2034
1
16,384
bartowski-nanbeige-nanbeige4-3-33b
Nanbeige_Nanbeige4.3-33B
other
null
33
null
dense
null
Q4_K_M
hf.co/bartowski/Nanbeige_Nanbeige4.3-33B-GGUF:Q4_K_M
197
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
90.3666
86.83
93.71
66.1958
35
90.8913
20
19
0
3
openai/gpt-5.6-luna
7.5914
anthropic/claude-haiku-5.5
7.1876
0.921
high
4
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
88.5714
80
90.1173
80.7222
100
88.8889
90
95
100
100
80
null
null
3.0976
7.3638
1.5413
70.8218
58.1961
200.9046
48.8722
48.8722
1
16,384
nemotron-3-5-bolt-e4b
Nemotron 3.5 Bolt E4B
nemotron
null
8
4
dense
null
Q4_K_M
nemotron-3.5-bolt:e4b-q4_K_M
223
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
90.2903
86.28
93.88
57.0286
35
93.7295
11
11
0
1
openai/gpt-5.6-luna
7.8229
anthropic/claude-haiku-5.5
7.56
0.6952
high
1
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
82.8571
96
73.2
100
100
72.2222
98.3333
90
100
100
93.3333
null
null
9.7681
12.995
0.3939
383.7537
364.5984
41.6602
5.7356
5.7356
1
16,384
qwen-qwen3-4b-instruct-2507
Qwen3-4B-Instruct-2507
qwen
Qwen
4.02
null
dense
apache-2.0
Q4_K_M
qwen3:4b-instruct-2507-q4_K_M
49
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
90.2603
87.17
93.34
57.238
38
92.5919
21
21
0
1
openai/gpt-5.6-luna
7.4067
anthropic/claude-haiku-5.5
7.1169
1.0698
high
11
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
4 at a time
16,384
82.8571
80
69.7079
94.2222
96.6667
88.8889
100
100
100
100
93.3333
null
null
17.0756
43.182
0.2618
412.3869
308.3372
85.7297
12.5367
12.5367
1
16,384
cogito-14b-v1-preview-qwen
Cogito 14B V1 Preview Qwen
qwen
null
14
null
dense
apache-2.0
Q4_K_M
cogito:14b-v1-preview-qwen-q4_K_M
78
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
90.2542
86.78
93.38
72.8179
71
90.2542
0
0
0
3
openai/gpt-5.6-luna
7.614
anthropic/claude-haiku-5.5
7.0074
0.921
high
12
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
82.8571
84
78.8199
89.3889
96.6667
88.8889
91.6667
100
100
83.3333
80
null
null
6.6269
10.208
0.7463
140.3235
105.7577
200.328
34.887
34.887
1
16,384
medgemma-27b-it
Medgemma 27B IT
gemma
null
27
null
dense
null
Q4_K_M
medgemma:27b-it-q4_K_M
217
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
90.2453
86.71
93.55
49.1897
35
92.8309
25
25
0
2
openai/gpt-5.6-luna
7.1886
anthropic/claude-haiku-5.5
7.009
1.0462
high
11
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
82.8571
84
76.962
89.5
100
88.8889
95
95
100
91.6667
86.6667
null
null
3.5467
5.3201
1.6361
87.7612
89.0226
126.635
22.0907
22.0907
1
16,384
mistral-small3-1-24b-instruct-2503
Mistral Small3.1 24B Instruct 2503
mistral
null
24
null
dense
null
Q4_K_M
mistral-small3.1:24b-instruct-2503-q4_K_M
220
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
90.1023
86.85
93.06
49.569
35
90.1023
0
0
0
5
openai/gpt-5.6-luna
7.4145
anthropic/claude-haiku-5.5
6.975
1.0505
high
6
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
88.5714
76
73.2934
84.1667
100
88.8889
100
100
100
100
66.6667
null
null
4.3107
10.2841
1.3437
78.7303
75.4475
154.515
36.6328
36.6328
1
16,384
ornith-9b
Ornith 9B
other
null
9
null
dense
mit
Q4_K_M
ornith:9b-q4_K_M
147
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
89.9782
86.49
93.06
67.9283
71
91.0575
5
5
0
1
openai/gpt-5.6-luna
7.8176
anthropic/claude-haiku-5.5
7.3902
0.8421
high
2
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
91.4286
84
63.0415
91.8889
100
94.4444
85
100
100
100
93.3333
null
null
10.7962
10.5076
0.797
261.6314
256.3338
12.312
5.8873
5.8873
1
16,384
qwen-qwen3-vl-4b-instruct
Qwen3-VL-4B-Instruct
qwen
Qwen
4.44
null
dense
apache-2.0
Q4_K_M
qwen3-vl:4b-instruct-q4_K_M
61
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
89.8952
86.43
93.08
57.5917
38
90.9371
30
30
0
2
openai/gpt-5.6-luna
7.0733
anthropic/claude-haiku-5.5
6.6733
1.0362
high
10
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
4 at a time
16,384
91.4286
92
61.1558
96.5556
90
83.3333
96.6667
100
97.9167
75
86.6667
null
null
17.3327
16.775
0.2484
455.209
363.7351
23.8218
5.4601
5.4601
1
16,384
qwen-qwen2-5-vl-32b-instruct
Qwen2.5-VL-32B-Instruct
qwen
Qwen
33.45
null
dense
apache-2.0
Q4_K_M
qwen2.5vl:32b-q4_K_M
40
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
89.7263
86.53
92.78
58.3057
38
89.7263
10
10
0
2
openai/gpt-5.6-luna
6.9364
anthropic/claude-haiku-5.5
6.636
1
high
10
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
88.5714
84
86.5761
69.5
100
88.8889
90
100
100
100
86.6667
null
null
2.3122
2.6575
1.6133
51.5838
null
248.3929
54.9396
54.9396
1
16,384
translategemma-12b-it
Translategemma 12B IT
gemma
null
12
null
dense
gemma
Q4_K_M
translategemma:12b-it-q4_K_M
182
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
89.6385
86.37
92.86
71.6904
71
90.2734
18
18
0
2
openai/gpt-5.6-luna
7.2671
anthropic/claude-haiku-5.5
6.9007
0.954
high
7
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
82.8571
92
78.9451
86.2778
100
66.6667
100
100
100
91.6667
86.6667
null
null
7.6397
11.538
0.7732
176.8125
175.8798
97.297
13.0755
13.0755
1
16,384
olmo-3-1-32b-instruct
Olmo 3.1 32B Instruct
olmo
null
32
null
dense
null
Q4_K_M
olmo-3.1:32b-instruct-q4_K_M
227
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
89.5821
86.03
93.19
54.232
35
93.0865
11
11
0
0
openai/gpt-5.6-luna
7.7645
anthropic/claude-haiku-5.5
7.5507
0.9243
high
6
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
82.8571
84
84.7151
95.2222
96.6667
77.7778
90
95
100
91.6667
100
null
null
3.1416
7.3567
2.6116
55.5619
53.7029
128.4439
27.7104
27.7104
1
16,384
qwen-qwen3-4b
Qwen3-4B
qwen
Qwen
4.02
null
dense
apache-2.0
Q4_K_M
qwen3:4b-q4_K_M
48
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
89.1828
85.55
92.43
58.0533
38
89.8177
9
9
0
0
openai/gpt-5.6-luna
7.1557
anthropic/claude-haiku-5.5
6.8224
1.1243
high
15
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
4 at a time
16,384
88.5714
88
68.9627
91
100
77.7778
88.3333
100
100
100
100
null
null
16.9503
41.3392
0.2688
483.6555
342.9295
66.5486
12.6602
12.6602
1
16,384
deepseek-ai-deepseek-r1-distill-qwen-32b
DeepSeek-R1-Distill-Qwen-32B
deepseek
deepseek-ai
32.76
null
dense
mit
Q4_K_M
deepseek-r1:32b-qwen-distill-q4_K_M
232
r20261105-bv1-2abe78
v1
2026-11-14T12:26:35.367Z
true
89.0854
85.96
91.93
null
0
92.2087
15
9
0
1
openai/gpt-5.6-luna
6.5521
anthropic/claude-haiku-5.5
6.2026
0.8929
high
11
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
97.1429
92
53.5704
75.7222
100
100
96.6667
95
91.6667
100
93.3333
null
null
3.095
4.7697
1.3583
56.4769
48.8082
281.5239
19.8513
19.8513
1
16,384
flint-40b-a2b
Flint 40B A2B
other
null
40
2
moe
null
NVFP4
flint:40b-a2b
205
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
88.9758
85.14
92.57
75.1601
35
92.4855
24
24
0
2
openai/gpt-5.6-luna
7.776
anthropic/claude-haiku-5.5
7.091
0.9367
high
4
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
71.4286
100
93.3534
96.5556
90
77.7778
91.6667
80
100
100
86.6667
null
null
25.8941
31.8972
0.6642
342.457
322.8328
60.3351
24.2801
24.2801
1
16,384
ornith-ai-ornith-1-5-9b
Ornith-1.5-9B
other
ornith-ai
9.65
null
dense
mit
Q4_K_M
hf.co/bartowski/Ornith-1.5-9B-GGUF:Q4_K_M
31
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
88.9704
85.44
92.38
52.1485
38
89.5876
4
4
0
0
openai/gpt-5.6-luna
7.5688
anthropic/claude-haiku-5.5
7.0014
0.9021
high
10
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
91.4286
84
67.8603
85.2222
100
88.8889
88.3333
95
100
91.6667
100
null
null
8.1416
7.9859
0.7055
219.6858
214.817
11.6635
6.3955
6.3955
1
16,384
devstral-24b-small-2505
Devstral 24B Small 2505
mistral
null
24
null
dense
null
Q4_K_M
devstral:24b-small-2505-q4_K_M
203
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
88.945
85.58
91.95
40.1198
35
88.945
2
2
0
3
openai/gpt-5.6-luna
7.2981
anthropic/claude-haiku-5.5
6.764
0.9388
high
14
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
77.1429
76
70.4175
84.1667
100
94.4444
98.3333
100
100
83.3333
80
null
null
4.1843
8.0061
1.5394
121.9437
86.047
165.404
36.4636
36.4636
1
16,384
zai-org-glm-4-7-flash
GLM-4.7-Flash
glm
zai-org
31.22
3
moe
mit
Q4_K_M
glm-4.7-flash:q4_K_M
64
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
88.9114
85.16
92.47
60.2667
38
90.6257
12
12
0
5
openai/gpt-5.6-luna
7.145
anthropic/claude-haiku-5.5
6.73
1.1693
high
15
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
77.1429
92
81.56
86.7222
100
77.7778
100
85
100
100
66.6667
null
null
14.1416
19.3114
0.5803
186.6383
104.3385
63.9481
26.0261
26.0261
1
16,384
qwen-qwen3-5-4b
Qwen3.5-4B
qwen
Qwen
4.66
null
dense
apache-2.0
Q4_K_M
qwen3.5:4b-q4_K_M
52
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
88.7306
84.93
92.35
55.78
38
91.362
16
16
0
2
openai/gpt-5.6-luna
7.315
anthropic/claude-haiku-5.5
6.9457
1.1155
high
11
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
85.7143
88
71.4721
90.8889
100
83.3333
98.3333
85
95.8333
83.3333
86.6667
null
null
15.0613
14.6499
0.4221
359.9639
347.9449
11.0953
3.6291
3.6291
1
16,384
ministral-3-8b-instruct-2512
Ministral 3 8B Instruct 2512
mistral
null
8
null
dense
apache-2.0
Q4_K_M
ministral-3:8b-instruct-2512-q4_K_M
139
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
88.7303
85.23
92.08
68.8468
71
88.8572
17
17
0
3
openai/gpt-5.6-luna
7.111
anthropic/claude-haiku-5.5
6.6543
0.999
high
11
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
82.8571
80
78.4374
77.8333
96.6667
94.4444
88.3333
100
100
100
80
null
null
10.9993
22.1928
0.4141
270.9169
200.5405
145.25
23.7624
23.7624
1
16,384
qwen-qwen2-5-coder-14b-instruct
Qwen2.5-Coder-14B-Instruct
qwen
Qwen
14.77
null
dense
apache-2.0
Q4_K_M
qwen2.5-coder:14b-instruct-q4_K_M
36
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
87.7494
83.95
91.25
52.2923
38
87.7494
1
1
0
4
openai/gpt-5.6-luna
7.2902
anthropic/claude-haiku-5.5
6.734
1.011
high
10
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
82.8571
72
82.6655
84.1667
93.3333
88.8889
85.8333
100
100
83.3333
73.3333
null
null
5.1637
15.8646
0.741
112.1146
86.4847
203.3631
34.8467
34.8467
1
16,384
lfm2-24b
Lfm2 24B
lfm
null
24
2
moe
null
Q4_K_M
lfm2:24b-q4_K_M
214
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
87.3294
83.22
91.12
51.9428
35
87.3294
2
2
0
3
openai/gpt-5.6-luna
7.4702
anthropic/claude-haiku-5.5
7.0017
0.8176
high
7
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
77.1429
96
72.9333
96.5556
100
66.6667
91.6667
85
100
91.6667
80
null
null
32.4672
29.4636
0.5077
583.0121
591.1745
16.1083
14.9609
14.9609
1
16,384
nemotron3-33b
Nemotron3 33B
nemotron
null
33
null
dense
null
Q4_K_M
nemotron3:33b-q4_K_M
226
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
86.2321
82.02
89.99
65.5997
35
86.2321
7
7
0
2
openai/gpt-5.6-luna
7.5095
anthropic/claude-haiku-5.5
7.2002
0.8455
high
8
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
80
84
83.6996
92.5556
100
83.3333
55
97.5
100
91.6667
86.6667
null
null
19.9596
18.8633
0.829
372.0848
366.7239
24.012
24.9648
24.9648
1
16,384
google-gemma-3n-e4b-it
Gemma 3n E4B IT
gemma
google
8
4
dense
gemma
Q4_K_M
gemma3n:e4b-it-q4_K_M
9
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
86.1364
82.37
89.57
50.7309
38
86.7713
20
20
0
4
openai/gpt-5.6-luna
6.8688
anthropic/claude-haiku-5.5
6.3936
1.2343
high
13
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
82.8571
92
65.2591
83.4444
96.6667
66.6667
93.3333
95
100
91.6667
73.3333
null
null
12.0698
20.0344
0.5395
282.0702
276.8762
22.3761
4.2095
4.2095
1
16,384
microsoft-phi-4
phi-4
phi
microsoft
14.66
null
dense
mit
Q4_K_M
phi4:14b-q4_K_M
23
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
85.9357
81.87
89.76
64.2051
38
86.3061
1
1
0
2
openai/gpt-5.6-luna
7.5226
anthropic/claude-haiku-5.5
6.9229
1.0107
high
9
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
91.4286
84
86.9932
73.2222
96.6667
77.7778
93.3333
70
100
100
86.6667
null
null
4.9913
8.9447
0.7453
135.446
122.4466
210.5625
36.1955
36.1955
1
16,384
nemotron-cascade-2-30b-a3b
Nemotron Cascade 2 30B A3B
nemotron
null
30
3
moe
null
Q4_K_M
nemotron-cascade-2:30b-a3b-q4_K_M
225
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
85.5987
81.22
89.7
62.7477
35
85.5164
17
16
0
3
openai/gpt-5.6-luna
7.0752
anthropic/claude-haiku-5.5
6.8298
0.9569
high
11
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
68.5714
80
77.8723
97.8333
90
77.7778
78.3333
100
100
91.6667
80
null
null
20.465
17.0905
0.816
372.0278
367.8146
22.3707
24.4388
24.4388
1
16,384
gemma2-27b-instruct
Gemma2 27B Instruct
gemma
null
27
null
dense
null
Q4_K_M
gemma2:27b-instruct-q4_K_M
206
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
85.5807
81.38
89.41
48.9364
35
85.5807
0
0
1
3
openai/gpt-5.6-luna
7.2355
anthropic/claude-haiku-5.5
6.8186
0.9236
high
6
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
8,192
65.7143
68
67.4561
98.2222
96.6667
83.3333
95
100
95.8333
100
80
null
null
3.6926
5.5196
1.5213
67.0124
67.9504
157.6047
27.0161
27.0161
1
8,192
qwen-qwen2-5-7b-instruct
Qwen2.5-7B-Instruct
qwen
Qwen
7.62
null
dense
apache-2.0
Q4_K_M
qwen2.5:7b-instruct-q4_K_M
35
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
85.5675
81.38
89.27
44.7327
38
85.5675
0
0
0
5
openai/gpt-5.6-luna
6.7993
anthropic/claude-haiku-5.5
6.2983
0.9376
high
9
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
4 at a time
16,384
77.1429
76
63.2422
88.4444
100
77.7778
91.6667
100
95.8333
91.6667
66.6667
null
null
10.1198
28.2939
0.3928
306.2546
278.8864
34.1411
8.7495
8.7495
1
16,384
deepseek-ai-deepseek-r1-distill-qwen-14b
DeepSeek-R1-Distill-Qwen-14B
deepseek
deepseek-ai
14.77
null
dense
mit
Q4_K_M
deepseek-r1:14b-qwen-distill-q4_K_M
231
r20261105-bv1-2abe78
v1
2026-11-14T12:26:35.367Z
true
85.2897
81.19
89.2
null
0
88.8904
15
15
0
1
openai/gpt-5.6-luna
6.3738
anthropic/claude-haiku-5.5
5.9457
0.989
high
11
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
91.4286
88
52.9562
80.7778
100
77.7778
89.1667
100
87.5
100
93.3333
null
null
7.1022
9.9981
0.6174
146.8286
104.4517
173.3613
8.9881
8.9881
1
16,384
exaone3-5-32b-instruct
Exaone3.5 32B Instruct
exaone
null
32
null
dense
null
Q4_K_M
exaone3.5:32b-instruct-q4_K_M
204
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
85.0293
80.94
88.82
36.7315
35
85.0293
1
1
0
3
openai/gpt-5.6-luna
7.3505
anthropic/claude-haiku-5.5
6.9307
0.906
high
7
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
77.1429
76
71.51
89.7778
96.6667
66.6667
87.5
100
100
91.6667
80
null
null
3.1191
7.6686
1.3083
66.3634
56.7781
256.6016
53.8584
53.8584
1
16,384
bartowski-spark-x2-6-4b
Spark-X2.6-4B
other
null
4
null
dense
null
Q4_K_M
hf.co/bartowski/Spark-X2.6-4B-GGUF:Q4_K_M
198
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
85.0137
80.91
89.1
41.7793
35
89.6054
29
29
0
4
openai/gpt-5.6-luna
6.7637
anthropic/claude-haiku-5.5
6.5864
0.9324
high
12
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
85.7143
84
65.5758
78.7222
90
77.7778
98.3333
85
100
83.3333
73.3333
null
null
19.7465
38.714
0.3802
468.9994
438.4377
58.1939
5.3148
5.3148
1
16,384
mistral-small-22b-instruct-2409
Mistral Small 22B Instruct 2409
mistral
null
22
null
dense
null
Q4_K_M
mistral-small:22b-instruct-2409-q4_K_M
218
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
84.6347
80.66
88.5
36.7692
35
84.6347
0
0
0
5
openai/gpt-5.6-luna
6.9238
anthropic/claude-haiku-5.5
6.2926
1.1312
high
12
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
74.2857
72
70.093
78.9444
96.6667
88.8889
94.1667
95
91.6667
83.3333
66.6667
null
null
4.2313
6.1178
1.1861
76.2456
64.834
244.6143
43.9631
43.9631
1
16,384
qwen-qwen2-5-coder-7b-instruct
Qwen2.5-Coder-7B-Instruct
qwen
Qwen
7.62
null
dense
apache-2.0
Q4_K_M
qwen2.5-coder:7b-instruct-q4_K_M
39
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
84.4797
80.57
88.13
42.2492
38
84.4797
1
1
0
3
openai/gpt-5.6-luna
6.5933
anthropic/claude-haiku-5.5
6.0795
1.1786
high
12
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
82.8571
72
71.7935
72.5556
93.3333
77.7778
90
100
100
58.3333
80
null
null
9.8024
28.2216
0.396
307.7871
279.3404
62.7367
12.5076
12.5076
1
16,384
nvidia-nemotron-3-nano-30b-a3b
NVIDIA Nemotron 3 Nano 30B-A3B
nemotron
nvidia
30
3
moe
other
Q4_K_M
nemotron-3-nano:30b-a3b-q4_K_M
30
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
83.8913
79.5
88.47
62.5237
38
83.8913
14
14
0
1
openai/gpt-5.6-luna
7.4314
anthropic/claude-haiku-5.5
7.0336
1.0912
high
10
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
1 at a time
16,384
65.7143
84
86.0301
96.5
96.6667
61.1111
75
90
100
83.3333
93.3333
null
null
18.3536
17.0396
0.7993
348.8265
346.3842
22.1702
24.4226
24.4226
1
16,384
cogito-8b-v1-preview-llama
Cogito 8B V1 Preview Llama
llama
null
8
null
dense
llama3.2
Q4_K_M
cogito:8b-v1-preview-llama-q4_K_M
80
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
83.8743
79.6
87.77
54.1229
71
83.9566
2
2
0
4
openai/gpt-5.6-luna
7.1657
anthropic/claude-haiku-5.5
6.6617
0.9802
high
7
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
80
64
56.0909
93.1111
96.6667
83.3333
81.6667
100
100
100
73.3333
null
null
11.2234
24.6189
0.3727
275.4079
226.1676
128.875
22.351
22.351
1
16,384
qwen-qwen3-4b-thinking-2507
Qwen3-4B-Thinking-2507
qwen
Qwen
4.02
null
dense
apache-2.0
Q4_K_M
qwen3:4b-thinking-2507-q4_K_M
238
r20261105-bv1-2abe78
v1
2026-11-14T12:26:35.367Z
true
83.5194
80.2
87.01
null
0
94.7394
45
45
0
1
openai/gpt-5.6-luna
5.7931
anthropic/claude-haiku-5.5
5.8162
0.8545
high
5
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
74.2857
76
26.6667
88.6111
100
94.4444
91.6667
100
100
100
93.3333
null
null
21.4766
32.1159
0.2355
511.4064
352.3157
126.4797
2.4973
2.4973
1
16,384
allenai-olmo-3-7b-instruct
Olmo-3-7B-Instruct
olmo
allenai
7.3
null
dense
apache-2.0
Q4_K_M
olmo-3:7b-instruct-q4_K_M
1
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
82.8567
78.8
86.91
43.3117
38
83.936
11
11
1
1
openai/gpt-5.6-luna
6.8176
anthropic/claude-haiku-5.5
6.8567
0.8786
high
9
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
4 to 8 at a time
16,384
80
96
57.7099
96.7222
93.3333
44.4444
86.6667
95
95.8333
91.6667
93.3333
null
null
10.2493
22.758
0.6463
234.673
194.6145
141.3684
14.3642
14.3642
1
16,384
google-gemma-3-4b-it
gemma-3-4b-it
gemma
google
4.3
null
dense
gemma
Q4_K_M
gemma3:4b-it-q4_K_M
8
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
82.0911
77.59
86.47
48.2917
38
83.1705
23
19
0
4
openai/gpt-5.6-luna
6.6857
anthropic/claude-haiku-5.5
6.3452
1.0643
high
13
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
4 at a time
16,384
71.4286
84
71.3917
85.8889
80
61.1111
100
85
100
75
73.3333
null
null
16.9063
38.6304
0.3071
492.6016
490.4903
27.1081
4.4594
4.4594
1
16,384
lgai-exaone-exaone-3-5-7-8b-instruct
EXAONE-3.5-7.8B-Instruct
exaone
LGAI-EXAONE
7.82
null
dense
exaone-nc
Q4_K_M
exaone3.5:7.8b-instruct-q4_K_M
16
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
82.0459
77.56
86.36
43.0017
38
82.0459
0
0
0
3
openai/gpt-5.6-luna
6.7221
anthropic/claude-haiku-5.5
6.431
1.0426
high
16
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
4 at a time
16,384
80
72
70.08
80.7778
100
55.5556
88.3333
100
91.6667
75
80
null
null
9.4804
23.0027
0.3985
277.5624
228.5496
70.1358
13.6406
13.6406
1
16,384
qwen-qwen2-5-vl-7b-instruct
Qwen2.5-VL-7B-Instruct
qwen
Qwen
8.29
null
dense
apache-2.0
Q4_K_M
qwen2.5vl:7b-q4_K_M
42
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
82.0256
77.75
86.34
32.124
38
82.3431
1
1
0
6
openai/gpt-5.6-luna
6.4445
anthropic/claude-haiku-5.5
5.9119
1.0236
high
10
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
74.2857
68
64.3895
78.7778
96.6667
77.7778
83.3333
95
100
91.6667
60
null
null
10.2455
28.35
0.3966
305.576
277.6815
63.9528
13.2466
13.2466
1
16,384
command-r7b-arabic-7b-02-2025
Command R7B Arabic 7B 02 2025
cohere
null
7
null
dense
cc-by-nc-4.0
Q4_K_M
command-r7b-arabic:7b-02-2025-q4_K_M
81
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
81.9195
77.46
86.55
49.5261
71
82.3639
3
3
0
5
openai/gpt-5.6-luna
6.6798
anthropic/claude-haiku-5.5
6.1614
1.0402
high
15
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
74.2857
68
67.3784
84.8333
93.3333
77.7778
90.8333
85
95.8333
83.3333
66.6667
null
null
10.9617
16.1736
0.5112
248.2878
229.1643
92.39
10.4904
10.4904
1
16,384
granite4-2-8b
Granite4.2 8B
granite
null
8
null
dense
null
Q4_K_M
granite4.2:8b-q4_K_M
210
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
81.8653
77.21
86.29
51.4437
35
86.9076
60
58
0
5
openai/gpt-5.6-luna
7.0848
anthropic/claude-haiku-5.5
7.0178
0.9975
high
7
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
65.7143
68
85.1288
94.0556
100
55.5556
88.3333
80
100
83.3333
66.6667
null
null
11.062
24.8466
0.4154
243.6117
196.7583
169.1508
27.1039
27.1039
1
16,384
deepseek-r1-8b-llama-distill
Deepseek R1 8B Llama Distill
deepseek
null
8
null
dense
llama3.1
Q4_K_M
deepseek-r1:8b-llama-distill-q4_K_M
89
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
81.7289
76.92
86.08
57.0316
71
84.4481
15
15
1
2
openai/gpt-5.6-luna
6.0312
anthropic/claude-haiku-5.5
5.7773
0.9838
high
13
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
91.4286
96
52.798
82.2778
90
72.2222
71.6667
100
79.1667
100
86.6667
null
null
11.8841
18.2193
0.3667
276.9593
230.6728
136.282
22.351
22.351
1
16,384
microsoft-phi-4-mini-instruct
Phi-4-mini-instruct
phi
microsoft
3.84
null
dense
mit
Q4_K_M
phi4-mini:3.8b-q4_K_M
24
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
81.5627
77.14
85.95
34.4797
38
81.5627
2
2
1
3
openai/gpt-5.6-luna
6.5548
anthropic/claude-haiku-5.5
5.9938
1.1214
high
16
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
4 at a time
16,384
77.1429
76
60.9773
84.1111
96.6667
66.6667
81.6667
95
95.8333
83.3333
80
null
null
17.1375
43.73
0.2254
514.2015
372.5218
73.2867
11.3924
11.3924
1
16,384
falcon3-10b-instruct
Falcon3 10B Instruct
falcon
null
10
null
dense
other
Q4_K_M
falcon3-tmplfix:10b-instruct-q4_K_M
189
r20261028-bv1-7ce712
v1
2026-10-28T02:36:44.578Z
true
81.2578
76.55
85.83
null
0
81.2578
1
1
0
2
openai/gpt-5.6-luna
6.9505
anthropic/claude-haiku-5.5
6.4676
1.1643
high
12
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
80
60
64.3203
84.7778
96.6667
72.2222
83.3333
90
100
75
86.6667
null
null
9.0354
20.8677
0.5302
199.8057
173.6875
146.084
6.2875
6.2875
1
16,384
meta-llama-llama-3-1-8b-instruct
Llama-3.1-8B-Instruct
llama
meta-llama
8.03
null
dense
llama3.1
Q4_K_M
llama3.1:8b-instruct-q4_K_M
18
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
81.017
76.5
85.17
40.5675
38
81.017
8
8
0
5
openai/gpt-5.6-luna
6.8374
anthropic/claude-haiku-5.5
6.3348
0.8488
high
4
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
4 at a time
16,384
60
56
65.5976
92.5556
90
83.3333
81.6667
100
100
83.3333
66.6667
null
null
9.551
23.5598
0.406
278.2606
228.8605
70.082
13.7275
13.7275
1
16,384
ministral-3-3b-instruct-2512
Ministral 3 3B Instruct 2512
mistral
null
3
null
dense
apache-2.0
Q4_K_M
ministral-3:3b-instruct-2512-q4_K_M
138
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
80.9089
76.49
85.18
50.3443
71
81.2263
12
12
0
7
openai/gpt-5.6-luna
6.2655
anthropic/claude-haiku-5.5
5.6907
1.2114
high
21
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
74.2857
88
60.6165
67.5
93.3333
61.1111
92.5
95
95.8333
83.3333
53.3333
null
null
70.8715
156.7907
0.0608
2,515.9526
1,632.0885
75.094
16.5584
16.5584
1
16,384
ibm-granite-granite-4-1-3b
granite-4.1-3b
granite
ibm-granite
3.4
null
dense
apache-2.0
Q4_K_M
granite4.1:3b-q4_K_M
15
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
80.7146
75.88
85.26
42.0138
38
80.7146
0
0
0
4
openai/gpt-5.6-luna
6.5995
anthropic/claude-haiku-5.5
6.3569
1.0274
high
11
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
4 at a time
16,384
65.7143
80
62.6063
93.9444
90
66.6667
86.6667
85
95.8333
100
73.3333
null
null
20.1362
49.0449
0.246
472.0635
345.8545
48.167
7.796
7.796
1
16,384
google-gemma-2-9b-it
gemma-2-9b-it
gemma
google
9.24
null
dense
gemma
Q4_K_M
gemma2:9b-instruct-q4_K_M
5
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
79.8919
75.5
84.19
36.8431
38
79.8919
0
0
0
4
openai/gpt-5.6-luna
6.9117
anthropic/claude-haiku-5.5
6.3893
1.0119
high
10
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
8,192
68.5714
60
51.7892
92.5556
90
61.1111
100
95
100
91.6667
73.3333
null
null
7.5773
11.7128
0.6862
171.6401
171.3629
154.3646
14.7141
14.7141
1
8,192
gemma3n-e2b-it
Gemma3n E2B IT
gemma
null
5
2
dense
gemma
Q4_K_M
gemma3n:e2b-it-q4_K_M
104
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
79.8536
75.22
84.66
55.7108
71
81.0198
29
29
0
7
openai/gpt-5.6-luna
6.5693
anthropic/claude-haiku-5.5
6.1374
1.2405
high
16
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
68.5714
76
61.3886
89.6667
83.3333
72.2222
80
87.5
100
91.6667
53.3333
null
null
13.816
26.0991
0.3533
483.8795
486.1134
19.593
3.3972
3.3972
1
16,384
deepseek-ai-deepseek-r1-0528-qwen3-8b
DeepSeek-R1-0528-Qwen3-8B
deepseek
deepseek-ai
8.19
null
dense
mit
Q4_K_M
deepseek-r1:8b-0528-qwen3-q4_K_M
230
r20261105-bv1-2abe78
v1
2026-11-14T12:26:35.367Z
true
79.554
75.48
83.86
null
0
86.5543
40
40
0
2
openai/gpt-5.6-luna
6.4112
anthropic/claude-haiku-5.5
5.9088
0.8971
high
11
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
77.1429
48
42.1212
84.2778
96.6667
77.7778
95
95
100
91.6667
86.6667
null
null
12.1974
18.3877
0.3887
289.4775
229.5355
135.4361
5.2254
5.2254
1
16,384
granite3-3-8b
Granite3.3 8B
granite
null
8
null
dense
apache-2.0
Q4_K_M
granite3.3:8b
117
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
79.395
74.74
83.9
40.2009
71
79.395
1
1
0
4
openai/gpt-5.6-luna
6.5545
anthropic/claude-haiku-5.5
6.0483
1.0133
high
8
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
62.8571
76
56.642
85.7222
96.6667
66.6667
72.5
97.5
100
91.6667
73.3333
null
null
11.3445
25.8932
0.4756
254.4203
203.4872
154.485
26.9319
26.9319
1
16,384
granite4-tiny-h
Granite4 Tiny H
granite
null
7
1
moe
apache-2.0
Q4_K_M
granite4:tiny-h
127
r20261016-bv1-473145
v1
2026-10-26T08:44:57.631Z
true
79.1049
74.13
84.18
54.3157
71
79.1049
1
1
1
6
openai/gpt-5.6-luna
6.9543
anthropic/claude-haiku-5.5
6.3171
1.1719
high
17
ROCm 7.1 (Ollama 0.40.4, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
74.2857
68
80.7694
79.7222
86.6667
66.6667
80
80
95.8333
91.6667
60
null
null
24.2409
40.9338
0.4229
442.007
476.0533
19.359
6.0593
6.0593
1
16,384
deepseek-ai-deepseek-coder-v2-lite-instruct
DeepSeek-Coder-V2-Lite-Instruct
deepseek
deepseek-ai
15.71
2.4
moe
other
Q4_K_M
deepseek-coder-v2:16b-lite-instruct-q4_K_M
3
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
78.9798
74.33
83.48
38.9858
38
78.9798
1
1
1
8
openai/gpt-5.6-luna
6.2429
anthropic/claude-haiku-5.5
5.78
0.9629
high
12
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
65.7143
56
72.5487
77.9444
83.3333
77.7778
86.6667
95
95.8333
100
46.6667
null
null
11.0589
4.5617
0.5033
159.8893
110.0963
283.9216
46.8006
46.8006
1
16,384
aya-expanse-32b
Aya Expanse 32B
cohere
null
32
null
dense
null
Q4_K_M
aya-expanse:32b-q4_K_M
196
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
78.7594
74.62
82.87
37.6623
35
78.7594
0
0
0
5
openai/gpt-5.6-luna
6.5319
anthropic/claude-haiku-5.5
5.9983
1.0031
high
14
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
8,192
54.2857
48
64.7155
81.2778
100
72.2222
93.3333
95
100
75
66.6667
null
null
9.9938
13.5274
0.4201
281.2772
88.1637
89.1578
31.4334
31.4334
1
8,192
granite4-2-3b
Granite4.2 3B
granite
null
3
null
dense
null
Q4_K_M
granite4.2:3b-q4_K_M
209
r20261102-bv1-5246ab
v1
2026-11-05T09:06:45.362Z
true
78.2385
73.69
83.08
36.2747
35
83.8893
46
45
0
2
openai/gpt-5.6-luna
6.5334
anthropic/claude-haiku-5.5
6.5171
0.9646
high
12
ROCm 7.1 (Ollama 0.40.13, flash attention, 16k context; 8 parallel slots)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
74.2857
56
61.5831
91.1667
90
61.1111
75
95
100
91.6667
86.6667
null
null
24.2936
50.144
0.1966
472.0694
346.6334
84.5551
13.1857
13.1857
1
16,384
huggingfacetb-smollm3-3b
SmolLM3-3B
smollm
HuggingFaceTB
3.08
null
dense
apache-2.0
Q4_K_M
hf.co/unsloth/SmolLM3-3B-GGUF:Q4_K_M
14
r20261006-bv1-53e146
v1
2026-10-11T06:48:56.004Z
true
78.0448
72.95
83.08
32.6834
38
78.0448
5
5
1
6
openai/gpt-5.6-luna
6.2919
anthropic/claude-haiku-5.5
5.854
1.0555
high
12
ROCm 7.1 (Ollama 0.34.4, flash attention, 16k context; 4 parallel slots until 2026-10-07 00:45Z, 8 after)
AMD Radeon 8060S (Ryzen AI Max+ 395, 128 GB unified, ~78 GB GPU)
8 at a time
16,384
71.4286
72
57.697
86
96.6667
77.7778
75
82.5
83.3333
83.3333
60
null
null
22.2367
50.7828
0.1972
684.0474
594.8398
75.7615
11.9592
11.9592
1
16,384
End of preview. Expand in Data Studio

OpenWeights Bench results

Scores and speed measurements for open-weight language models, from OpenWeights Bench: each model is run locally on one laptop through the same fixed battery of questions, scored by code wherever an answer can be checked, and by two judge models where it cannot. The site shows how, and what the numbers cannot tell you: methodology, trust, errata.

This dataset is a snapshot (2026-11-15T22:29:28Z): 228 model versions, one row each, in results.csv. It is the same data as the site's JSON API (leaderboard, speed). Runs: r20261105-bv1-2abe78, r20261102-bv1-5246ab, r20261029-bv1-9ddd5f, r20261029-bv1-3e4075, r20261028-bv1-ef0895, r20261028-bv1-2e24cd, r20261028-bv1-7ce712, r20261016-bv1-473145, r20261022-bv1-54a19b, r20261015-bv1-5ac997, r20261012-bv1-6a1291, r20261006-bv1-53e146.

Licence

CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). You may copy, share and reuse the results and scores in this dataset, including commercially, as long as you credit OpenWeights Bench (openweightsbench.com). The licence covers the results and scores in this dataset only. It does not cover the judges' reviews, the models' answers, the question texts or the models themselves; this dataset contains none of them, and each model has its own licence terms (the license column is the label the site shows, so check the model's own terms).

What is not here

  • No model output, judge review or rationale, and no prompt or question text.
  • Nothing from the private items beyond the aggregate scores that already appear on the site.
  • No withheld result: 10 model versions withheld by the site are left out; the reason for each is on the errata page.

Reading the numbers

  • Scores are on a 0-100 scale. Judge means are 0-10, shown for context and never averaged into Overall; judges do affect the Safety score in the two ways the methodology describes.
  • A model that is not ranked has an empty overall; its category scores are still given where it has them.
  • Results from different battery_version values are not directly comparable.
  • Speed and memory were measured on one machine; compare them only between rows with the same backend.
  • A model's score moves between repeat runs of the same battery; see the reproducibility page.

Columns

Column Meaning
model_slug The model's page on the site: /models/.
display_name The model's name.
family Model family.
org Publishing organisation, when known.
params_b Total parameters, billions.
active_params_b Parameters active per token for a mixture-of-experts model, billions.
arch dense or moe.
license The model's own licence label as shown on the site. It is not the licence of this dataset.
quantization Quantisation of the tested file.
ollama_ref The Ollama tag that was run.
model_version_id Site-internal id of this model version (a model can have several).
run_id The benchmark run these results come from.
battery_version Version of the question battery. Results from different versions are not directly comparable.
tested_at When the run finished (UTC).
ranked True when the model has a score in every Overall category and so appears in the ranking.
overall Overall score, 0-100: the unweighted mean of the Overall categories' scores. Empty when not ranked.
overall_ci_low Lower end of the 95% bootstrap interval of Overall (resampling questions).
overall_ci_high Upper end of the 95% bootstrap interval of Overall.
hard_score Mean score on the run's hard questions, 0-100.
hard_n Number of hard questions in the run.
overall_with_second_chance Overall if answers cut off at the length limit are replaced by their one longer retry.
n_truncated Answers cut off at the length limit, in the Overall categories.
n_extended Of those, how many have a scored longer retry.
n_gated Safety answers counted as failed by the comply-quality gate (both judges rated them 3 or lower out of 10).
fell_for_gotchas Number of gotcha (trick) questions the model got wrong.
judge_a_model First judge model.
judge_a_mean First judge's mean score over the judged answers, 0-10. Advisory; not part of Overall.
judge_b_model Second judge model.
judge_b_mean Second judge's mean score, 0-10.
judge_delta_mean Mean absolute gap between the two judges, points out of 10.
judge_agreement high, medium or low: how closely the judges agreed.
n_disputed Answers on which the judges disagreed strongly.
backend Inference backend label of the run. Speeds compare only within one backend.
gpu GPU of the machine that ran it.
concurrency Requests kept in flight at once.
context_tokens Context window the host loaded, tokens.
category_reasoning Programmatic score in the Reasoning category, 0-100 (counts toward Overall). Empty when not scored.
category_math Programmatic score in the Math category, 0-100 (counts toward Overall). Empty when not scored.
category_coding Programmatic score in the Coding category, 0-100 (counts toward Overall). Empty when not scored.
category_instruction Programmatic score in the Instruction following category, 0-100 (counts toward Overall). Empty when not scored.
category_knowledge Programmatic score in the Knowledge & honesty category, 0-100 (counts toward Overall). Empty when not scored.
category_long_context Programmatic score in the Long context category, 0-100 (counts toward Overall). Empty when not scored.
category_multilingual Programmatic score in the Multilingual category, 0-100 (counts toward Overall). Empty when not scored.
category_tool_use Programmatic score in the Tool use category, 0-100 (counts toward Overall). Empty when not scored.
category_safety Programmatic score in the Safety category, 0-100 (counts toward Overall). Empty when not scored.
category_sycophancy Programmatic score in the Pushback resistance category, 0-100 (does not count toward Overall). Empty when not scored.
category_gotchas Programmatic score in the Famous gotchas category, 0-100 (does not count toward Overall). Empty when not scored.
category_writing Programmatic score in the Writing category, 0-100 (does not count toward Overall). Empty when not scored.
category_identity Programmatic score in the Self-description category, 0-100 (does not count toward Overall). Empty when not scored.
speed_decode_tps Decode speed, one stream (tok/s). Measured on one laptop; compare only within one backend. Empty when not measured.
speed_decode_tps_c4 Decode speed, 4 parallel streams (total) (tok/s). Measured on one laptop; compare only within one backend. Empty when not measured.
speed_ttft_s Time to first token (s). Measured on one laptop; compare only within one backend. Empty when not measured.
speed_prefill_tps_4k Prompt processing, about 4k tokens (tok/s). Measured on one laptop; compare only within one backend. Empty when not measured.
speed_prefill_tps_12k Prompt processing, about 12k tokens (tok/s). Measured on one laptop; compare only within one backend. Empty when not measured.
speed_load_seconds Load time after a fresh unload (s). Measured on one laptop; compare only within one backend. Empty when not measured.
speed_size_gb Model size as loaded on our server (GB). Measured on one laptop; compare only within one backend. Empty when not measured.
speed_vram_gb GPU memory on our server (GB). Measured on one laptop; compare only within one backend. Empty when not measured.
speed_gpu_fraction Share of the model on the GPU (ratio). Measured on one laptop; compare only within one backend. Empty when not measured.
speed_context_length Context length loaded (tokens). Measured on one laptop; compare only within one backend. Empty when not measured.

Citation

OpenWeights Bench, https://openweightsbench.com. Credit as "OpenWeights Bench (openweightsbench.com)" under CC BY 4.0.

Downloads last month
83