TokenCostEstimate bill →
← All benchmarks·Source: Epoch AI ↗

GPQA Diamond leaderboard

Current leader: GPT-5.4 Pro at 92.8 (% accuracy).

Cost vs quality

Each dot is a model. The dashed line is the value frontier — top-left is the best trade-off (high score, low cost).

Closed model Open-weight model

Full leaderboard

#ModelScore% accuracyOutput $/1MReleased
1
GPT-5.4 Pro
92.8$180.00Mar 2026
2
Gemini 3.1 Pro
92.13$12.00Feb 2026
3
GPT-5.5
92$30.00Apr 2026
4
GPT-5.5 Pro
91.9$180.00Apr 2026
5
Claude Opus 5
91.84$50.00Jul 2026
6
GPT-5.6 Sol
91.33$30.00Jul 2026
7
Grok 4.5
91.25$6.00Jul 2026
8
GPT-5.6 Terra
91.08$7.50Jul 2026
9
GPT-5.4
91.07$15.00Mar 2026
10
Kimi K3OPEN
90.82$15.00Jul 2026
11
Gemini 3.5 Flash
90.4$9.00—
12
Gemini 3 Pro
90.15—Nov 2025
13
GLM-5.2OPEN
89.14$2.35Jun 2026
14
Qwen3.7-Max
88.8$4.42May 2026
15
GPT-5.6 Luna
88.8$3.00Jul 2026
16
GPT-5.2
88.53—Dec 2025
17
Claude Opus 4.8
88.05$50.00May 2026
18
Kimi K2.6OPEN
87.71$2.72Apr 2026
19
Claude Opus 4.6
87.37$25.00Feb 2026
20
Claude Sonnet 5
87.37$10.00Jun 2026
21
Claude Opus 4.7
86.87$150.00Apr 2026
22
Muse Spark
86.4—Apr 2026
23
DeepSeek-V4-ProOPEN
86.2$0.87Apr 2026
24
Kimi K2.7 CodeOPEN
86.03$3.50Jun 2026
25
Grok 4.20
85.77$2.50Feb 2026
26
Qwen 3.6 Max (Preview)
85.44$6.16Apr 2026
27
Grok 4.3 Beta
85.1—Apr 2026
28
GLM-5OPEN
83.76$2.55Feb 2026
29
GPT-5.1
83.5—Nov 2025
30
Kimi K2.5OPEN
83.47$2.85Feb 2026
31
Qwen 3.6 Plus
83.16$1.95Apr 2026
32
Claude Sonnet 4.6
83.16$15.00Feb 2026
33
Grok 4
82.67—Sep 2025
34
GPT-5
81.57—Aug 2025
35
Claude Opus 4.5
81.4—Nov 2025
36
GLM-5.1
80.64$3.04Apr 2026
37
Gemini 2.5 Pro (Jun 2025)
80.39—Jun 2025
38
Qwen 3.6 Flash
79.23$1.13Apr 2026
39
Kimi K2 ThinkingOPEN
78.96—Nov 2025
40
Qwen 3.5 Plus (hosted 397B-A17B)
78.87$1.80Feb 2026
41
Gemini 2.5 Pro (Mar 2025)
78.45—Mar 2025
42
Qwen 3.5 Flash (hosted 35B-A3B)
78.45$0.26Feb 2026
43
GPT-5.4 Mini
78.11$4.50Mar 2026
44
DeepSeek-V3.2OPEN
77.9—Dec 2025
45
GLM-4.7OPEN
77.78—Dec 2025
46
Gemini 3 Flash
77.61—Dec 2025
47
Claude Sonnet 4.5
76.43—Sep 2025
48
o3
75.76—Dec 2024
49
Qwen3-235B-A22B-Thinking (Jul 2025)OPEN
73.4—Jul 2025
50
Claude 3.7 Sonnet
72.98—Feb 2025
51
o4-mini
72.81—Apr 2025
52
Claude Sonnet 4
72.25—May 2025
53
GPT-5.4 Nano
71.29$1.25Mar 2026
54
Claude Opus 4.1
69.7—Aug 2025
55
o3-mini
69.36—Jan 2025
56
o1
69.02—Dec 2024
57
DeepSeek-R1 (May 2025)OPEN
68.43—May 2025
58
Grok-3 mini
68.35—Feb 2025
59
Claude Opus 4
68.35—May 2025
60
Grok 3
67.68—Feb 2025
61
gpt-oss-120bOPEN
67.68—Aug 2025
62
GPT-5 mini
66.67—Aug 2025
63
Qwen3-Max
63.47$3.90Jan 2026
64
DeepSeek-R1OPEN
62.29—Jan 2025
65
Claude Haiku 4.5
61.62—Oct 2025
66
Qwen3-235B-A22BOPEN
60.94—Apr 2025
67
GPT-5 nano
59.26—Aug 2025
68
GPT-4.5
58.25—Feb 2025
69
DeepSeek-V3 (Mar 2025)OPEN
56.82—Mar 2025
70
Llama 4 MaverickOPEN
55.98—Apr 2025
71
GPT-4.1
55.89—Apr 2025
72
Gemini 2.5 Pro (May 2025)
55.56—May 2025
73
GPT-4.1 mini
54.46—Apr 2025
74
Gemini 2.0 Pro
54.21—Dec 2024
75
Gemini 2.0 Flash (Feb 2025)
52.19—Dec 2024
76
o1-mini
49.83—Sep 2024
77
Mistral Medium 3
46.04—May 2025
78
Gemini 1.5 Pro (Sept 2024)
42.97—Sep 2024
79
Gemini 2.0 Flash Thinking (Jan 2025)
42.76—Jan 2025
80
DeepSeek-V3OPEN
42.05—Dec 2024
81
Qwen2.5-Max
41.5—Jan 2025
82
Phi-4OPEN
41.41—Apr 2025
83
Claude 3.5 Sonnet (October 2024)
40.4—Oct 2024
84
Claude 3.5 Sonnet
38.72—Jun 2024
85
Grok-2 (Dec 2024)
38.38—Aug 2024
86
Llama 4 ScoutOPEN
35.77—Apr 2025
87
Mistral Large 2 (Nov 2024)
35.1—Jul 2024
88
Llama 3.1-405BOPEN
34.55—Jul 2024
89
o1-preview
33.75—Dec 2024
90
GPT-4o (Aug 2024)
32.28—Aug 2024
91
Qwen2.5-72BOPEN
32.2—Sep 2024
92
Mistral Large 2 (Jul 2024)
32.03—Jul 2024
93
GPT-4.1 nano
31.9—Apr 2025
94
GPT-4o (May 2024)
31.86—May 2024
95
Gemma 3 27BOPEN
31.82—Mar 2025
96
Magistral Small 1.1OPEN
31.23—Jun 2025
97
GPT-4o (Nov 2024)
30.51—Nov 2024
98
Mistral Small 3.1OPEN
29.97—Mar 2025
99
Llama 3.3 70BOPEN
29.92—Dec 2024
100
Gemini 1.5 Flash (Sep 2024)
29.76—May 2024
101
Claude 3 Opus
29.55—Mar 2024
102
GPT-4 Turbo (Apr 2024)
28.79—Apr 2024
103
Gemini 1.5 Pro (May 2024)
27.82—May 2024
104
Llama 3.1-70BOPEN
25.59—Jul 2024
105
Llama 3.2 90BOPEN
21.38—Sep 2024
106
Qwen2-72BOPEN
21.04—Jun 2024
107
Claude 3 Sonnet
20.79—Mar 2024
108
Llama 3-70BOPEN
20.75—Apr 2024
109
Gemini 1.5 Flash (May 2024)
20.5—May 2024
110
Mistral Large
18.35—Feb 2024
111
Claude 3.5 Haiku
17.51—Oct 2024
112
GPT-4o mini
16.96—Jul 2024
113
Gemma 2 27BOPEN
15.32—Jun 2024
114
Claude 3 Haiku
15.07—Mar 2024
115
GPT-4 (Mar 2023)
14.31—Mar 2023
116
Claude 2
12.88—Jul 2023
117
Mixtral 8x22BOPEN
12.08—Apr 2024
118
Gemini 1.0 Pro
11.95—Dec 2023
119
Claude 2.1
10.61—Nov 2023
120
GPT-4 (Jun 2023)
7.53—Jun 2023
121
Mixtral 8x7BOPEN
7.45—Dec 2023
122
Mistral NeMoOPEN
6.52—Jul 2024
123
GPT-3.5 Turbo (Nov 2023)
4.04—Jun 2023
124
phi-3-medium 14BOPEN
3.45—Apr 2024
125
Gemma 2 9BOPEN
3.28—Jun 2024
126
GPT-3.5 Turbo (Jan 2024)
2.9—Jun 2023
127
Llama 2-70BOPEN
1.77—Jul 2023
128
Llama 3-8BOPEN
1.43—Apr 2024
129
Llama 3.1-8BOPEN
1.26—Jul 2024
130
Yi-34BOPEN
0—Nov 2023

Scores from Epoch AI, updated —. Cost is joined from our daily API pricing where a model matches; models we don't price still appear with their score. Always verify on the benchmark's own site before relying on a number.