MMMU Pro (Vals) · Open Analysis

#	Model	Effort	Score	Cost / task	Harness
1	Fable 5.1	max	90.6%	$0.166	–
2	Opus 5	max	89.9%	$0.069	–
3	Fable 5	max	89.3%	$0.161	–
4	Gemini 3.8 Flash	high	89.1%	$0.0045	–
5	Gemini 3.7 Flash	high	89.0%	$0.014	–
6	GPT-5.6 Sol	max	88.8%	$0.056	–
7	Gemini 3.6 Flash	high	88.4%	$0.019	–
8	Gemini 3.5 Flash	high	88.3%	$0.025	–
9	GPT-5.5	xhigh	88.3%	$0.043	–
10	Gemini 3.1 Pro	high	88.2%	$0.064	–
11	Kimi K3	max	88.2%	$0.059	–
12	Qwen 3.8 Max	max	88.0%	$0.016	–
13	Gemini 3 Flash	high	87.6%	$0.014	–
14	Gemini 3 Pro	high	87.5%	$0.010	–
15	GPT-5.4	xhigh	87.5%	$0.047	–
16	Muse Spark	–	87.4%	$0.0017	–
17	GPT-5.2	xhigh	86.7%	$0.061	–
18	Muse Spark 1.1	xhigh	86.6%	$0.014	–
19	Opus 4.8	max	86.6%	$0.121	–
20	GPT-5.6 Terra	max	86.5%	$0.016	–
21	Kimi K2.6	max	86.3%	$0.035	–
22	Muse Spark 1.2	xhigh	86.1%	$0.018	–
23	GLM 5.3 Flash	max	86.0%	$0.0010	–
24	Opus 4.7	max	85.5%	$0.170	–
25	GPT-5.6 Luna	max	85.0%	$0.0064	–
26	Kimi K2.5	max	84.3%	$0.020	–
27	Qwen 3.6 Plus	max	84.2%	$0.017	–
28	Qwen 3.8 27B	xhigh	83.9%	$0.013	–
29	Opus 4.6	max	83.9%	$0.116	–
30	Gemini 3.5 Flash-Lite	high	83.6%	$0.0066	–
31	Sonnet 4.6	max	83.6%	$0.144	–
32	Grok 4.20	xhigh	83.5%	$0.027	–
33	GPT-5.1	high	83.2%	$0.0028	–
34	Grok 4.3	xhigh	83.1%	$0.022	–
35	Sonnet 5	max	83.0%	$0.033	–
36	Opus 4.5	high	82.9%	$0.127	–
37	Gemini 3.1 Flash-Lite	high	82.5%	$0.0032	–
38	Qwen 3.5 Flash	max	81.9%	$0.0026	–
39	GPT-5	high	81.5%	$0.033	–
40	Gemini 2.5 Pro	high	81.3%	$0.0078	–
41	MiniMax M3	max	81.2%	$0.0094	–
42	Gemini 2.5 Flash	medium	80.8%	$0.0019	–
43	o3	high	80.4%	$0.015	–
44	MiMo V2.5	xhigh	80.0%	$0.0014	–
45	o4 mini	high	79.7%	$0.011	–
46	Sonnet 4.5	high	79.3%	$0.058	–
47	GPT-5.4 Mini	xhigh	79.2%	$0.0071	–
48	GPT-5 Mini	high	78.9%	$0.0072	–
49	Inkling	max	78.5%	$0.031	–
50	Opus 4.1	high	77.5%	$0.169	–
51	o1	high	77.4%	$0.146	–
52	Grok 4	high	76.3%	$0.098	–
53	Gemini 2.5 Flash Lite	none	75.4%	$0.0004	–
54	Sonnet 3.7	medium	75.1%	$0.051	–
55	Sonnet 4	high	74.9%	$0.051	–
56	GPT-5.4 Nano	high	73.6%	$0.0008	–
57	Opus 4	medium	73.3%	$0.062	–
58	Grok 4 Fast	none	72.8%	$0.0029	–
59	Grok 4.1 Fast	high	72.7%	$0.0006	–
60	GPT-4.1	high	72.4%	$0.0080	–
61	Llama 4 Maverick	medium	71.7%	$0.0011	–
62	GPT-5 Nano	high	70.9%	$0.0026	–
63	GPT-4.1 Mini	high	70.5%	$0.0018	–
64	Gemini 2.0 Flash	medium	69.8%	$0.0004	–
65	Claude 3.5 Sonnet Latest	–	68.8%	$0.010	–
66	Mistral Large 3	none	66.2%	$0.0015	–
67	Gemini 1.5 Pro	none	65.5%	$0.0018	–
68	Magistral Small 1.2	–	65.2%	$0.0044	–
69	Magistral Medium 1.2	medium	64.6%	$0.021	–
70	GPT-4o	high	64.0%	$0.0063	–
71	Mistral Medium 3.1	none	63.0%	$0.0016	–
72	Grok 4.5	high	61.8%	$0.030	–
73	Mistral Small 3.1	none	60.1%	$0.0003	–
74	Llama 4 Scout	medium	58.8%	$0.0008	–
75	Grok 2	none	57.3%	$0.0054	–
76	Gemini 1.5 Flash	none	57.2%	$0.0001	–
77	GPT-4o mini	high	56.6%	$0.0004	–
78	GPT-4.1 Nano	high	55.1%	$0.0006	–
79	Llama 3.2	–	48.1%	$0.0083	–
80	Haiku 4.5	xhigh	46.1%	$0.027	–
81	Qwen 3.5 Plus	none	22.8%	$0.015	–