AA-MMMU-Pro leaderboard

A display-only Artificial Analysis MMMU-Pro score.

Models ranked by their published Artificial Analysis MMMU-Pro value

Rank	Model	Provider	Image + text question answering
1	GPT-6 Astra	OpenAI	86.9
2	Gemini 3.8 Flash	Google	85.6
3	Gemini 3.7 Flash	Google	85.5
4	Claude Opus 5	Anthropic	84.7
5	Gemini 3.5 Flash	Google	84.3
6	GPT-5.6 Sol	OpenAI	83.4
7	Gemini 3.6 Flash	Google	83.2
8	Qwen3.8 Max Preview	Alibaba	82.8
9	Gemini 3.1 Pro	Google	82.4
10	GPT-5.6 Terra	OpenAI	80.7
11	Kimi K3	Moonshot AI	80.5
11	Muse Spark	Meta	80.5
11	Qwen3.7 Plus	Alibaba	80.5
14	Grok 4.5	xAI	80.4
15	Gemini 3 Pro	Google	80.2
16	GPT-5.5	OpenAI	79.9
17	Qwen3.8-Flash-Next	Alibaba	79.8
18	Kimi K2.6	Moonshot AI	79.4
19	APApodex 1.1	Apodex	79.2
19	APApodex 1.1 Mini	Apodex	79.2
21	Gemini 3.5 Flash-Lite	Google	79.0
21	Ling 3.0 Flash VL	InclusionAI	79.0
23	Claude Opus 4.7 (Adaptive)	Anthropic	78.8
24	Gemini 3 Flash	Google	78.6
24	GPT-5.6 Luna	OpenAI	78.6
24	MiniMax M3	MiniMax	78.6
27	GPT-5.3 Codex	OpenAI	78.5
28	GPT-5.4	OpenAI	78.4
29	Grok 4.3	xAI	78.1
30	Qwen3.6 Plus	Alibaba	78.0
31	Claude Sonnet 5	Anthropic	77.3
32	DeepSeek V4.1 Flash	DeepSeek	77.0
33	Claude Opus 4.7	Anthropic	76.4
34	GPT-5.2-Codex	OpenAI	76.3
34	Qwen3.8-27B	Alibaba	76.3
36	GPT-5.1	OpenAI	75.5
37	Claude Opus 4.6 (Adaptive)	Anthropic	75.4
37	Kimi K2.5	Moonshot AI	75.4
37	Kimi K2.5 (Reasoning)	Moonshot AI	75.4
40	Step 3.7 Flash	StepFun	75.3
41	Qwen3.5-122B-A10B	Alibaba	75.0
41	Qwen3.5-27B	Alibaba	75.0
41	Qwen3.6-35B-A3B	Alibaba	75.0
44	Gemini 2.5 Pro	Google	74.9
45	Qwen3.6-27B	Alibaba	74.6
46	GPT-5 (medium)	OpenAI	74.3
46	Muse Glimmer 30B	Meta	74.3
48	GPT-5 (high)	OpenAI	74.2
49	Claude Opus 4.5 Thinking	Anthropic	74.0
49	TMInkling-Small	Thinking Machines Lab	74.0
51	TMInkling	Thinking Machines Lab	73.5
52	Gemma 4 31B	Google	73.4
53	GPT-5.4 mini	OpenAI	73.3
54	GLM-5V-Turbo	Z.AI	72.8
55	Qwen3.5-35B-A3B	Alibaba	72.7
56	Claude Opus 4.6	Anthropic	72.5
56	GPT-5.1-Codex	OpenAI	72.5
56	GPT-5.1-Codex-Max	OpenAI	72.5
59	Claude Opus 4.5	Anthropic	71.2
60	Claude Sonnet 4.6	Anthropic	70.6
61	o3	OpenAI	70.1
62	MiMo-V2-Omni	Xiaomi	69.9
63	Gemma 4 12B	Google	69.7
64	Gemma 4 26B A4B	Google	69.2
65	Grok 4	xAI	68.8
66	Claude 4.1 Opus Thinking	Anthropic	67.9
67	Gemini 2.5 Flash	Google	65.5
68	GPT-5.4 nano	OpenAI	65.4
69	Mistral Medium 3.5 128B	Mistral	64.9
70	Grok 4.1 Fast (Reasoning)	xAI	63.3
71	Command A+	Cohere	63.2
72	Claude 4 Sonnet	Anthropic	62.4
73	Llama 4 Maverick	Meta	62.1
74	Grok 4 Fast (Reasoning)	xAI	61.8
75	GPT-4.1	OpenAI	61.2
76	Qwen3-Omni-30B-A3B-Thinking	Alibaba	60.2
77	GPT-4.1 mini	OpenAI	58.7
78	Mistral Small 4	Mistral	56.8
78	Mistral Small 4 (Reasoning)	Mistral	56.8
80	Mistral Large 3	Mistral	55.7
81	Qwen3-Omni-30B-A3B-Instruct	Alibaba	55.5
82	Gemini 1.5 Pro	Google	55.0
83	Nemotron 3 Nano Omni 30B A3B	NVIDIA	53.2
84	Mistral Medium 3	Mistral	53.0
85	Llama 4 Scout	Meta	52.9
86	Qwen3.5 397B	Alibaba	52.7
86	Qwen3.5 397B (Reasoning)	Alibaba	52.7
88	Gemma 4 E4B	Google	51.4
89	Grok 4.1 Fast	xAI	48.4
90	Gemma 3 27B	Google	48.0
91	Gemma 4 E2B	Google	44.6
92	Nova Pro	Amazon	44.3
93	GPT-4o mini	OpenAI	41.5
94	GPT-4.1 nano	OpenAI	40.1
95	Claude 3 Haiku	Anthropic	30.8
96	LFM2.5-VL-1.6B-Extract	LiquidAI	26.5
97	Phi-4 Multimodal Instruct	Microsoft	14.5

Evidence key: Observed

Rows are ordered by the value Artificial Analysis MMMU-Pro Benchmark Leaderboard published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.