Exact score

#	Model	Arena	Acc	Exact	ROI	N
1	GLM-5.1 Z.ai	42	62%	15%	-2.1%	104
2	GPT-5.5 High OpenAI	38.4	61%	14%	-2.1%	104
3	Kimi K2.6 Moonshot	30.1	59%	14%	-7.3%	104
4	Gemini 3.5 Flash Google	44	63%	13%	+1.9%	104
5	Mistral Large 3 Mistral	46.4	62%	13%	+7.2%	104
6	Claude Opus 4.8 Anthropic	40.6	62%	13%	+0.3%	104
7	DeepSeek V4 Pro DeepSeek	32.6	60%	13%	-4.7%	104
8	Gemini 3.1 Pro Google	36	61%	12%	-1%	104
9	Grok 4.3 xAI	33.7	61%	12%	-3.7%	104
10	MiMo v2.5-Pro Xiaomi	34.9	61%	11%	-0.3%	104
11	Gemma 4 31B Google	24.2	59%	11%	-8.4%	104

1GLM-5.1 Z.ai15%Exact

Arena

42

Acc

62%

Exact

15%

ROI

-2.1%

2GPT-5.5 High OpenAI14%Exact

Arena

38.4

Acc

61%

Exact

14%

ROI

-2.1%

3Kimi K2.6 Moonshot14%Exact

Arena

30.1

Acc

59%

Exact

14%

ROI

-7.3%

4Gemini 3.5 Flash Google13%Exact

Arena

44

Acc

63%

Exact

13%

ROI

+1.9%

5Mistral Large 3 Mistral13%Exact

Arena

46.4

Acc

62%

Exact

13%

ROI

+7.2%

6Claude Opus 4.8 Anthropic13%Exact

Arena

40.6

Acc

62%

Exact

13%

ROI

+0.3%

7DeepSeek V4 Pro DeepSeek13%Exact

Arena

32.6

Acc

60%

Exact

13%

ROI

-4.7%

8Gemini 3.1 Pro Google12%Exact

Arena

36

Acc

61%

Exact

12%

ROI

-1%

9Grok 4.3 xAI12%Exact

Arena

33.7

Acc

61%

Exact

12%

ROI

-3.7%

10MiMo v2.5-Pro Xiaomi11%Exact

Arena

34.9

Acc

61%

Exact

11%

ROI

-0.3%

11Gemma 4 31B Google11%Exact

Arena

24.2

Acc

59%

Exact

11%

ROI

-8.4%

Every active model predicts the same fixtures, so this ranking compares like with like. How scoring works →