MMMU · validation leaderboard (October 2026)

Model	MMMU validation	Per task	Price	Speed	Context
GPT-5.1OpenAI	85.4unstated	—	$1.25 / $10	—	400K
o3OpenAI	82.9unstated	—	$2 / $8	—	200K
Claude Opus 4.5Anthropic	80.7unstated	—	$5 / $25	—	200K
dDots.vlm1.instdots-studioOpen	80.1	—		—	164K
Gemini 2.5 Pro Preview (May' 25)Google	79.6unstated	—		—	1M
Qwen3 VL 235B A22BAlibaba (Qwen)	78.7	—	$0.53 / $2.66	—	256K
o1OpenAI	78.2unstated	—	$15 / $60	—	200K
Claude Sonnet 4.5Anthropic	77.8unstated	—	$3 / $15	—	1M
iIntern S1internlmOpen	77.7	—		—	66K
Skywork R1V3 38BSkyworkOpen	76	—		—	33K
Skywork R1V2 38BSkyworkOpen	73.6	—		—	41K
Llama 4 MaverickMetaOpen	73.4unstated	—	$0.19 / $0.65	—	1M
GPT-4oOpenAI	70.7unstated	—	$2.50 / $10	—	128K
QVQ 72B PreviewAlibaba (Qwen)Open	70.3	—	$0.25 / $0.50	—	32K
Llama 4 ScoutMetaOpen	69.4unstated	—	$0.10 / $0.30	—	10M
Skywork R1V 38BSkyworkOpen	69	—		—	131K
Nova PremierAmazon	68unstated	—	$2.50 / $12.50	—	1M
MiMo VL 7B RLXiaomiOpen	66.7	—		—	128K
Gemma 3 27BGoogleOpen	64.9unstated	—	$0.08 / $0.16	—	131K
MiMo VL 7B SFTXiaomiOpen	64.6	—		—	128K
OInternVL3 8BOpenGVLabOpen	62.7	—		—	33K
Nova ProAmazon	62	—	$0.80 / $3.20	—	300K
Llama 3.2 90BMeta	60.3	—	$0.72 / $0.72	—	—
Gemma 3 12BGoogleOpen	59.6unstated	—	$0.05 / $0.10	—	131K
GPT-4o MiniOpenAI	59.4unstated	—	$0.15 / $0.60	—	128K
Qianfan VL 70BBaiduOpen	58.3	—		—	33K
Eagle2.5 8BNVIDIAOpen	55.8	—		—	—
rAriarhymes-aiOpen	54.9	—		—	66K
Llama 3.2 11BMeta	50.7	—	$0.16 / $0.16	—	—
OMiniCPM V 2 6OpenBMBOpen	49.8	—		—	—
Gemma 3 4BGoogleOpen	48.8unstated	—	$0.04 / $0.05	—	131K
OInternVL3 2BOpenGVLabOpen	48.6	—		—	33K
Gemini 1.0 ProGoogle	47.9	—		—	—
Qianfan VL 8BBaiduOpen	47.1	—		—	33K
Qianfan VL 3BBaiduOpen	46.4	—		—	33K
OMiniCPM Llama3 V 2 5OpenBMBOpen	45.8	—	$0.34 / $0.34	—	8K
OInternVL3 1BOpenGVLabOpen	43.4	—		—	33K
FFlashVL 2B DynamicFlashVLOpen	42.9	—		—	33K
rReka EdgerekaaiOpen	42.8unstated	—	$0.10 / $0.10	—	16K
OMiniCPM-VOpenBMB	37.2	—		—	—
AOvis2 1BATH-MaaSOpen	36	—		—	33K

Best for Vision & Documents