Views
No views yet
1Benchmark Accuracy Time(s)
2--------------------------------------------------------------
3MMLU 90.0% 145.3
4MMLU_PRO 73.0% 65.8
5HELLASWAG 94.0% 64.6
6TRUTHFULQA 91.0% 44.6
7ARC_CHALLENGE 91.0% 41.2
8WINOGRANDE 85.0% 39.1
9GSM8K 96.0% 308.7
10MATHQA 59.0% 44.7
11
12
13Model: Qwen3.6-27B-Architect-Polaris2-Fable-B-F451-mxfp4-mlx
14Benchmark Accuracy Time(s)
15--------------------------------------------------------------
16MMLU 88.0% 116.4
17MMLU_PRO 77.0% 59
18HELLASWAG 93.0% 62.2
19TRUTHFULQA 90.0% 43.1
20ARC_CHALLENGE 92.0% 39.5
21WINOGRANDE 82.0% 37.4
22GSM8K 96.0% 338.4
23MATHQA 52.0% 44.8
24
25
26Model: Qwen3.6-27B-Miraculix-mxfp8-mlx
27Benchmark Accuracy Time(s)
28--------------------------------------------------------------
29MMLU 82.0% 127.1
30MMLU_PRO 73.0% 63.4
31HELLASWAG 93.0% 65.5
32TRUTHFULQA 91.0% 48.3
33ARC_CHALLENGE 92.0% 45.7
34WINOGRANDE 83.0% 43.9
35GSM8K 96.0% 344.7
36MATHQA 54.0% 54
37
38Model: Qwen3.6-27B-Fable-Fusion-711-BigBubba-717-mxfp8-mlx
39Benchmark Accuracy Time(s)
40--------------------------------------------------------------
41MMLU 89.0% 158.2
42MMLU_PRO 75.0% 74.2
43HELLASWAG 93.0% 72.9
44TRUTHFULQA 93.0% 52.7
45ARC_CHALLENGE 92.0% 49.1
46WINOGRANDE 83.0% 47
47GSM8K 97.0% 454.8
48MATHQA 55.0% 60.9
49
50
51Model: Qwen3.6-27B-ElCHUD-mxfp8-mlx
52Benchmark Accuracy Time(s)
53--------------------------------------------------------------
54MMLU 88.0% 171
55MMLU_PRO 75.0% 78
56HELLASWAG 93.0% 74.2
57TRUTHFULQA 92.0% 53
58ARC_CHALLENGE 92.0% 49.5
59WINOGRANDE 84.0% 47
60GSM8K 96.0% 447.8
61MATHQA 55.0% 55.1
62
63Model: Qwen3.6-27B-Akka-mxfp8-mlx
64Benchmark Accuracy Time(s)
65--------------------------------------------------------------
66MMLU 84.0% 132.5
67MMLU_PRO 72.0% 75.1
68HELLASWAG 92.0% 74.4
69TRUTHFULQA 92.0% 53.4
70ARC_CHALLENGE 91.0% 50.9
71WINOGRANDE 83.0% 49.9
72GSM8K 97.0% 425.7
73MATHQA 53.0% 57