MMLU Pro (Vals) · Open Analysis

#	Model	Effort	Score	Cost / task	Harness
1	Fable 5.1	max	92.4%	$0.098	–
2	Opus 5	max	91.6%	$0.028	–
3	Fable 5	max	91.5%	$0.082	–
4	Gemini 3.1 Pro	high	91.0%	$0.028	–
5	Gemini 3.8 Flash	high	90.2%	$0.022	–
6	Gemini 3.7 Flash	high	90.1%	$0.011	–
7	Gemini 3 Pro	high	90.1%	$0.037	–
8	Opus 4.7	max	89.9%	$0.062	–
9	Opus 4.8	max	89.6%	$0.061	–
10	Gemini 3.5 Flash	high	89.5%	$0.025	–
11	Grok 4.6	high	89.4%	$0.015	–
12	Qwen 3.7 Max	max	89.3%	$0.017	–
13	Gemini 3.6 Flash	high	89.3%	$0.013	–
14	Grok 4.5	high	89.2%	$0.015	–
15	Opus 4.6	max	89.1%	$0.052	–
16	GPT-5.6 Sol	max	89.1%	$0.032	–
17	Muse Spark 1.1	xhigh	88.7%	$0.011	–
18	Qwen 3.8 Max	max	88.6%	$0.019	–
19	Gemini 3 Flash	high	88.6%	$0.019	–
20	Muse Spark 1.2	xhigh	88.3%	$0.011	–
21	GPT-5.5	xhigh	88.1%	$0.028	–
22	Kimi K3	max	88.0%	$0.021	–
23	Opus 4.1	high	87.9%	$0.124	–
24	Qwen 3.6 Plus	max	87.7%	$0.0088	–
25	Kimi K2.6	max	87.6%	$0.018	–
26	Sonnet 5	max	87.5%	$0.043	–
27	GPT-5.4	xhigh	87.5%	$0.051	–
28	Sonnet 4.5	high	87.4%	$0.040	–
29	Sonnet 4.6	max	87.3%	$0.066	–
30	Muse Spark	–	87.3%	$0.0017	–
31	Opus 4.5	high	87.3%	$0.215	–
32	DeepSeek V4	max	87.2%	$0.027	–
33	Qwen 3.5 Plus	none	87.2%	$0.0089	–
34	MiniMax M2.1	–	87.0%	$0.0032	–
35	DeepSeek V4 Pro 0813	max	87.0%	$0.013	–
36	GLM 5.1	max	86.9%	$0.012	–
37	GLM 5.3	max	86.8%	$0.017	–
38	GLM 5.2	max	86.7%	$0.014	–
39	GPT-5.6 Terra	xhigh	86.7%	$0.0080	–
40	GPT-5	high	86.5%	$0.024	–
41	GPT-5.1	high	86.4%	$0.019	–
42	Inkling	max	86.3%	$0.040	–
43	Grok 4.20	xhigh	86.3%	$0.013	–
44	Gemini 3.1 Flash-Lite	high	86.2%	$0.0011	–
45	GPT-5.2	xhigh	86.2%	$0.037	–
46	DeepSeek V4 Flash 0731	high	86.2%	$0.0027	–
47	Opus 4	medium	86.2%	$0.050	–
48	GLM 5.3 Flash	max	86.1%	$0.0011	–
49	GPT-5.6 Luna	max	86.0%	$0.0028	–
50	GLM 5	high	86.0%	$0.012	–
51	Kimi K2.5	max	85.9%	$0.0079	–
52	Grok 4.3	high	85.8%	$0.0095	–
53	Gemini 3.5 Flash-Lite	high	85.8%	$0.0045	–
54	Nemotron 3 Ultra 550B A55B	max	85.8%	–	–
55	o3	high	85.6%	$0.012	–
56	Inkling Small	xhigh	85.6%	$0.0057	–
57	Grok 4	high	85.3%	$0.050	–
58	Qwen 3 Max	max	85.0%	$0.027	–
59	DeepSeek V3.2	high	84.9%	$0.0011	–
60	MiMo V2.5 Pro	xhigh	84.6%	$0.0021	–
61	GPT-5.4 Mini	xhigh	84.6%	$0.0037	–
62	Qwen 3.8 27B	xhigh	84.3%	$0.011	–
63	MiniMax M3	max	84.2%	$0.0048	–
64	Grok 4.1 Fast	high	84.2%	$0.0011	–
65	Qwen 3.5 Flash	max	84.1%	$0.0017	–
66	Gemini 2.5 Pro	high	84.1%	$0.0090	–
67	Sonnet 4	high	83.9%	$0.046	–
68	Gemini 2.5 Flash	medium	83.7%	$0.0015	–
69	o1	high	83.5%	$0.116	–
70	DeepSeek R1	xhigh	83.2%	$0.012	–
71	MiMo V2.5	xhigh	82.9%	$0.0008	–
72	GLM 4.7	none	82.7%	$0.0080	–
73	Sonnet 3.7	medium	82.7%	$0.040	–
74	GPT-5 Mini	high	82.2%	$0.0052	–
75	GLM 4.6	none	82.2%	$0.0066	–
76	Ling 3.0 Flash	–	82.0%	$0.0009	–
77	Grok 3 Mini	high	81.4%	$0.0018	–
78	Qwen 3 235B	–	81.2%	$0.0023	–
79	GLM 4.5	xhigh	81.2%	$0.0098	–
80	Kimi K2	medium	81.1%	$0.0021	–
81	o4 mini	high	80.6%	$0.0055	–
82	GPT-4.1	high	80.5%	$0.0071	–
83	MiniMax M2.7	high	80.4%	$0.0039	–
84	MiniMax M2.5	high	80.1%	$0.0054	–
85	Grok 3	none	79.9%	$0.017	–
86	Mistral Large 3	none	79.8%	$0.0015	–
87	Grok 4 Fast	none	79.7%	$0.0017	–
88	DeepSeek V3	none	79.5%	$0.0015	–
89	Llama 4 Maverick	medium	79.4%	$0.0007	–
90	gpt-oss-120b	high	79.2%	$0.0014	–
91	Gemini 2.5 Flash Lite	none	79.1%	$0.0003	–
92	Haiku 4.5	xhigh	78.7%	$0.018	–
93	o3 mini	high	78.7%	$0.013	–
94	Claude 3.5 Sonnet Latest	–	78.4%	$0.0086	–
95	Gemini 2.0 Flash	medium	77.4%	$0.0004	–
96	GPT-4.1 Mini	high	77.2%	$0.0013	–
97	GPT-5.4 Nano	high	77.2%	$0.0004	–
98	GPT-5 Nano	high	76.1%	$0.0019	–
99	Grok 2	none	75.5%	$0.0060	–
100	Mistral Medium 3.5	high	75.3%	$0.042	–
101	Gemini 1.5 Pro	none	75.3%	$0.0026	–
102	Mistral Medium 3.1	none	75.3%	$0.0002	–
103	GPT-4o	high	74.1%	$0.0072	–
104	gpt-oss-20b	high	71.6%	$0.0006	–
105	Llama 3.3	none	70.8%	–	–
106	Mistral Large 11	–	69.7%	$0.0050	–
107	Llama 4 Scout	medium	69.6%	$0.0005	–
108	Command A	none	69.2%	$0.0079	–
109	Laguna XS.2	–	69.0%	$0.0006	–
110	Laguna M.1	–	68.8%	$0.0018	–
111	Magistral Medium 1.2	medium	68.7%	$0.015	–
112	Mistral Small 3.1	none	66.0%	$0.0002	–
113	Gemini 1.5 Flash	none	65.6%	$0.0002	–
114	Mistral Small 02	–	64.4%	$0.0005	–
115	Haiku 3.5	medium	64.1%	$0.0022	–
116	GPT-4.1 Nano	high	63.5%	$0.0003	–
117	GPT-4o mini	high	62.7%	$0.0004	–
118	Magistral Small 1.2	–	62.1%	$0.0036	–
119	Jamba 1.6 Large	none	49.8%	$0.0046	–
120	Command-R+ (Apr '24)	none	44.0%	$0.0057	–
121	Jamba 1.6 Mini	none	30.3%	$0.0004	–