MMMU Leaderboard

Multimodal Understanding benchmark testing vision-language models on expert-level tasks.

Data from LayerLens

As of November 21, 2026, the top-scoring model on MMMU is Gemini 3.1 Pro Preview at 83.8%, followed by GPT-5.5 at 83.0% and Gemini 3.5 Flash at 83.0%. 83 models have been evaluated on this benchmark.

Last updated: November 21, 2026

8 Ways to Use Fewer Tokens

Models

83

Best Score

83.8

Average

61.3

Std Dev

20.7

Categories
Multimodal
Sponsored The Grid | Spot Priced LLM API
The Grid | Spot Priced LLM API — Why pay list price when providers will bid for your API usage? Up to 80% off.
View spot prices
Provider
Model
Input $/M
Output $/M
MMMU
Actions
$1.000
$6.000
83.8
$5.000
$30.000
83.0
$1.500
$9.000
83.0
$0.750
$3.750
82.9
$0.500
$3.000
82.8
$0.500
$3.000
81.7
$2.000
$6.000
81.6
$0.040
$0.078
81.4
$0.490
$2.500
79.3
$0.490
$2.500
79.3
$0.230
$0.960
79.3
$1.250
$10.000
79.1
$0.260
$0.900
78.1
$0.200
$1.200
77.7
$0.195
$0.900
77.6
$0.030
$0.130
77.4
$5.000
$25.000
77.3
$5.000
$25.000
77.3
$0.188
$1.125
77.0
$0.065
$0.260
76.8
$0.216
$0.647
76.7
$5.000
$25.000
76.6
$0.065
$0.750
76.4
$2.000
$8.000
76.3
$5.000
$25.000
76.3
$5.000
$25.000
76.3
$0.050
$0.700
75.9
$5.000
$25.000
75.4
$0.125
$1.000
75.3
$3.000
$15.000
75.3
$0.300
$2.500
74.4
$0.080
$0.120
74.1
$0.100
$0.100
74.1
$0.300
$0.900
74.1
$0.250
$1.500
73.7
$3.000
$15.000
72.9
$0.050
$0.190
71.8
$3.000
$15.000
71.7
$0.300
$0.900
71.1
$0.300
$0.900
71.1
$0.025
$0.200
70.7
$2.000
$8.000
69.3
$0.200
$0.880
68.2
$1.000
$5.000
65.2
$1.000
$5.000
65.2
$1.500
$7.500
64.4
$0.500
$1.500
63.0
$1.750
$14.000
62.3
$1.250
$7.500
61.6
$0.625
$5.000
60.7
$2.500
$10.000
59.2
$2.000
$6.000
58.8
$0.440
$2.200
58.7
$0.440
$2.200
58.1
$0.780
$3.900
57.4
$0.075
$0.200
55.9
$0.600
$1.920
55.9
$0.600
$1.920
55.9
$0.375
$2.250
55.7
$0.080
$0.160
53.9
$0.300
$2.500
53.0
$0.300
$2.500
53.0
$0.064
$0.224
52.1
$0.064
$0.224
52.1
$0.800
$3.200
50.1
$0.120
$0.500
49.0
$0.120
$0.500
49.0
$0.014
$0.028
48.2
$0.060
$0.240
45.7
$0.100
$0.625
42.4
$0.100
$0.300
42.3
$0.070
$0.140
39.7
$2.500
$10.000
37.9
$0.100
$0.400
34.1
$0.100
$0.400
34.1
$0.200
$0.600
31.3
$0.390
$0.900
21.8
$0.320
$1.400
16.1
$0.320
$1.400
16.1
$0.260
$1.560
13.1
$0.270
$1.000
-
$0.270
$1.000
-
$0.750
$3.750
-

Pricing from OpenRouter. Benchmarks from Artificial Analysis.

About MMMU

Multimodal Understanding benchmark testing vision-language models on expert-level tasks.

This leaderboard shows all models with MMMU benchmark scores, ranked from highest to lowest. Pricing data is included to help you compare performance against cost.

Frequently Asked Questions

Multimodal Understanding benchmark testing vision-language models on expert-level tasks.
As of November 21, 2026, Gemini 3.1 Pro Preview leads the MMMU leaderboard with a score of 83.8. Rankings change as new models are released and evaluated.
Currently 83 models have been evaluated on MMMU, with an average score of 61.3 and standard deviation of 20.7.
Benchmark scores are updated when new evaluations are published by our data sources (Artificial Analysis and LayerLens). Pricing data is refreshed daily from OpenRouter.