BBH Leaderboard

Big-Bench Hard — challenging subset of BIG-Bench focusing on tasks where language models previously underperformed.

Data from LayerLens

As of November 21, 2026, the top-scoring model on BBH is Gemini 3.1 Pro Preview at 96.1%, followed by Qwen3.7 Max at 95.8% and Qwen3.8 Max at 95.5%. 145 models have been evaluated on this benchmark.

Last updated: November 21, 2026

8 Ways to Use Fewer Tokens

Models

145

Best Score

96.1

Average

78.9

Std Dev

19.3

Categories
Reasoning and Logic
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
BBH
Actions
$1.000
$6.000
96.1
$1.475
$4.425
95.8
$2.000
$6.000
95.5
$1.027
$6.162
95.2
$5.000
$25.000
94.6
$3.000
$15.000
94.4
$3.000
$15.000
94.3
$0.600
$1.920
94.3
$0.600
$1.920
94.3
$5.000
$25.000
94.1
$5.000
$30.000
94.1
$0.320
$1.280
94.0
$0.750
$3.750
94.0
$0.250
$0.900
93.5
$0.899
$2.825
93.3
$2.000
$6.000
93.3
$0.500
$3.000
93.1
$0.750
$3.750
92.7
$0.260
$0.900
92.7
$0.780
$3.900
92.6
$0.500
$3.000
92.4
$0.195
$0.900
92.4
$2.000
$10.000
92.4
$0.400
$1.760
92.1
$0.040
$0.078
92.0
$0.490
$2.500
91.9
$0.490
$2.500
91.9
$5.000
$25.000
91.8
$5.000
$25.000
91.8
$0.100
$0.100
91.8
$1.500
$9.000
91.7
$0.188
$1.125
91.6
$0.007
$0.950
91.6
$1.200
$4.000
91.5
$0.030
$0.130
91.5
$0.050
$0.700
91.3
$0.000
$0.000
91.2
$5.000
$25.000
91.1
$1.000
$4.050
91.1
$1.250
$10.000
90.9
$0.065
$0.750
90.8
$0.065
$0.260
90.6
$0.600
$3.000
90.6
$0.260
$1.560
90.6
$3.000
$15.000
90.3
$0.050
$0.190
90.2
$0.450
$0.900
90.2
$0.000
$0.000
90.1
$0.200
$1.200
89.7
$2.000
$8.000
89.6
$0.120
$0.500
89.5
$0.120
$0.500
89.5
$0.390
$0.900
89.4
$1.000
$10.000
88.7
$0.200
$0.880
88.7
$0.300
$0.900
88.6
$0.450
$1.200
88.6
$0.209
$0.310
88.5
$0.300
$0.900
88.5
$0.320
$1.400
88.4
$0.320
$1.400
88.4
$0.209
$0.310
88.2
$0.270
$0.410
88.0
$0.500
$2.150
87.9
$0.125
$1.000
87.7
$0.250
$1.500
87.7
$0.216
$0.647
87.6
$0.200
$0.900
87.5
$0.550
$2.200
87.1
$0.230
$0.960
87.1
$0.600
$2.200
87.0
$0.650
$3.250
87.0
$0.025
$0.200
86.9
$0.600
$2.500
86.7
$0.300
$2.500
86.2
$0.064
$0.224
85.8
$0.064
$0.224
85.8
$0.430
$1.750
85.5
$0.430
$1.750
85.5
$0.600
$2.500
85.4
$0.270
$0.950
85.4
$0.080
$0.200
84.6
$0.080
$0.200
84.6
$0.017
$0.085
84.4
$0.030
$0.100
84.3
$0.900
$0.900
84.2
$0.900
$0.900
84.2
$0.800
$3.200
83.9
$0.090
$0.900
82.3
$0.130
$0.850
82.2
$1.500
$7.500
82.0
$1.000
$5.000
82.0
$1.000
$5.000
82.0
$0.000
$0.000
81.9
$0.200
$0.600
81.8
$0.060
$0.240
81.8
$0.060
$0.240
81.8
$0.440
$2.200
81.5
$0.500
$1.500
81.3
$0.440
$2.200
80.5
$0.020
$0.096
80.1
$0.250
$0.950
79.7
$0.250
$0.950
79.7
$2.500
$10.000
79.6
$0.100
$0.300
79.3
$2.500
$10.000
76.7
$0.080
$0.160
76.6
$0.075
$0.200
76.5
$1.250
$7.500
75.3
$0.075
$0.400
75.2
$0.000
$0.000
74.3
$0.288
$0.320
72.4
$0.070
$0.280
71.9
$0.014
$0.028
69.8
$0.625
$5.000
69.8
$1.750
$14.000
67.6
$1.000
$3.000
67.2
$1.000
$3.000
67.2
$0.000
$0.000
64.2
$0.300
$0.900
62.9
$0.300
$2.500
61.6
$0.300
$2.500
61.6
$0.035
$0.140
61.5
$0.270
$1.000
60.5
$0.270
$1.000
60.5
$0.060
$0.240
59.0
$0.375
$2.250
58.0
$0.100
$0.320
56.6
$0.100
$0.400
56.0
$0.100
$0.400
56.0
$0.340
$0.390
55.9
$0.270
$0.410
52.8
$0.020
$0.040
46.1
$0.100
$0.625
45.6
$0.100
$0.280
43.8
$2.000
$6.000
41.2
$0.030
$0.050
36.0
$0.750
$0.800
35.8
$0.070
$0.140
34.5
$0.200
$0.200
28.7
$0.040
$0.040
18.4
$0.300
$0.300
17.1
$0.660
$0.800
11.4
$0.020
$0.020
8.7
$0.900
$0.900
2.9

Pricing from OpenRouter. Benchmarks from Artificial Analysis.

About BBH

Big-Bench Hard — challenging subset of BIG-Bench focusing on tasks where language models previously underperformed.

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

Frequently Asked Questions

Big-Bench Hard — challenging subset of BIG-Bench focusing on tasks where language models previously underperformed.
As of November 21, 2026, Gemini 3.1 Pro Preview leads the BBH leaderboard with a score of 96.1. Rankings change as new models are released and evaluated.
Currently 145 models have been evaluated on BBH, with an average score of 78.9 and standard deviation of 19.3.
Benchmark scores are updated when new evaluations are published by our data sources (Artificial Analysis and LayerLens). Pricing data is refreshed daily from OpenRouter.