Skip to main content
benchlm.ai

Artificial Analysis Humanity's Last Exam (AA-HLE)

A display-only Artificial Analysis Humanity's Last Exam score.

Benchmark score on AA-HLE — June 2, 2026

BenchLM mirrors the published score view for AA-HLE. Claude Opus 4.8 leads the public snapshot at 45.7% , followed by Gemini 3.1 Pro (44.7%) and GPT-5.5 (44.3%). BenchLM does not use these results to rank models overall.

1Closed

Claude Opus 4.8

Anthropic

45.7%
Overall 95Context 1M
2Closed

Gemini 3.1 Pro

Google

44.7%
Overall 92Context 1M
3Closed

GPT-5.5

OpenAI

44.3%
Overall 91Context 1M
124 modelsKnowledgeCurrentDisplay onlyUpdated June 2, 2026

The published AA-HLE snapshot is tightly clustered at the top: Claude Opus 4.8 sits at 45.7%, while the third row is only 1.4 points behind. The broader top-10 spread is 8.5 points, so many of the published scores sit in a relatively narrow band.

124 models have been evaluated on AA-HLE. The benchmark falls in the Knowledge category. This category carries a 12% weight in BenchLM.ai's overall scoring system. AA-HLE is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About AA-HLE

Year

2026

Tasks

Expert-level questions

Format

Accuracy

Difficulty

Frontier expert reasoning

BenchLM stores the Artificial Analysis HLE result separately from the weighted HLE lane so AA refreshes remain display-only.

Artificial Analysis Humanity's Last Exam Benchmark Leaderboard

BenchLM freshness & provenance

Version

AA-HLE 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

Benchmark score table (124 models)

1
Claude Opus 4.8
45.7%
2
Gemini 3.1 Pro
44.7%
3
GPT-5.5
44.3%
4
GPT-5.4
41.6%
5
Gemini 3.5 Flash
41.0%
6
GPT-5.3 Codex
39.9%
7
Muse Spark
39.9%
8
Claude Opus 4.7 (Adaptive)
39.6%
9
Qwen3.7 Max
38.1%
10
Gemini 3 Pro
37.2%
11
Claude Opus 4.6 (Adaptive)
36.7%
12
DeepSeek V4 Pro (Max)
35.9%
13
Kimi K2.6
35.9%
14
GPT-5.2
35.4%
15
Grok 4.3
35.0%
16
MiMo-V2.5-Pro
33.8%
17
DeepSeek V4 Pro (High)
33.5%
18
GPT-5.2-Codex
33.5%
19
DeepSeek V4 Flash (Max)
32.1%
20
Claude Opus 4.7
31.2%
21
Kimi K2.5 (Reasoning)
29.4%
22
Kimi K2.5
29.4%
23
Qwen 3.6 Max (preview)
28.9%
24
Claude Opus 4.5 Thinking
28.4%
25
MiMo-V2-Pro
28.3%
26
MiniMax M2.7
28.1%
27
GLM-5.1
28.0%
28
DeepSeek V4 Flash (High)
27.8%
29
Qwen3.5 397B (Reasoning)
27.3%
30
GLM-5
27.2%
31
GPT-5.4 mini
26.6%
32
GPT-5.1
26.5%
33
GPT-5 (high)
26.5%
34
GPT-5.4 nano
26.5%
35
Qwen3.6 Plus
25.7%
36
Hy3 Preview
25.5%
37
GLM-5-Turbo
25.4%
38
GLM-4.7
25.1%
39
Grok 4
23.9%
40
GPT-5 (medium)
23.5%
41
GPT-5.1-Codex-Max
23.4%
42
Qwen3.5-122B-A10B
23.4%
43
GPT-5.1-Codex
23.4%
44
Gemma 4 31B
22.7%
45
Qwen3.5-27B
22.2%
46
Qwen3.6-27B
21.6%
47
Gemini 2.5 Pro
21.1%
48
Qwen3.6-35B-A3B
20.2%
49
o3
20.0%
50
MiMo-V2-Omni
19.9%
51
Step 3.7 Flash
19.9%
52
Qwen3.5-35B-A3B
19.7%
53
Qwen3.5 397B
18.8%
54
Claude Opus 4.6
18.6%
55
GPT-OSS 120B
18.5%
56
Gemma 4 26B A4B
18.3%
57
Grok 4.1 Fast (Reasoning)
17.6%
58
Grok 4 Fast (Reasoning)
17.0%
59
Gemini 3.1 Flash-Lite
16.2%
60
GLM-5V-Turbo
15.8%
61
DeepSeek-R1
14.9%
62
Trinity-Large-Preview
14.7%
63
Trinity-Large-Thinking
14.7%
64
Gemini 3 Flash
14.1%
65
Claude Sonnet 4.6
13.2%
66
K-Exaone
13.1%
67
DeepSeek V3.1 (Reasoning)
13.0%
68
Claude Opus 4.5
12.9%
69
Mistral Medium 3.5 128B
12.8%
70
Claude 4.1 Opus Thinking
11.9%
71
Command A+
11.4%
72
Qwen3 Max
11.1%
73
DeepSeek V3.2
10.5%
74
Sarvam 105B
10.1%
75
GPT-OSS 20B
9.8%
76
Mistral Small 4 (Reasoning)
9.5%
77
Mistral Small 4
9.5%
78
o3-mini
8.7%
79
Nemotron Ultra 253B
8.1%
80
MiMo-V2-Flash
8.0%
81
o1
7.7%
82
Grok Code Fast 1
7.5%
83
Kimi K2
7.0%
84
Sarvam 30B
7.0%
85
GLM-4.5-Air
6.8%
86
Granite-4.0-H-350M
6.4%
87
DeepSeek V3.1
6.3%
88
Ling 2.6 Flash
6.2%
89
Exaone 4.0 1.2B
5.8%
90
Granite-4.0-350M
5.7%
91
DeepSeek R1 Distill Qwen 32B
5.5%
92
Nemotron 3 Nano Omni 30B A3B
5.3%
93
GLM-4.6
5.2%
94
Gemini 2.5 Flash
5.1%
95
Granite-4.0-1B
5.1%
96
Grok 4.1 Fast
5.0%
97
Granite-4.0-H-1B
5.0%
98
Gemini 1.5 Pro
4.9%
99
Exaone 4.0 32B
4.9%
100
Llama 4 Maverick
4.8%
101
Gemma 4 E2B
4.8%
102
Gemma 3 27B
4.7%
103
GPT-4.1
4.6%
104
GPT-4.1 mini
4.6%
105
Nemotron 3 Nano 30B
4.6%
106
Gemini 1.0 Pro
4.6%
107
Llama 4 Scout
4.3%
108
Mistral Medium 3
4.3%
109
Llama 3.1 405B
4.2%
110
Mistral Large 3
4.1%
111
Phi-4
4.1%
112
Claude 4 Sonnet
4.0%
113
GPT-4o mini
4.0%
114
Mistral Large 2
4.0%
115
GPT-4.1 nano
3.9%
116
Claude 3 Haiku
3.9%
117
Qwen2.5 Coder 32B Instruct
3.8%
118
Solar Pro 2
3.8%
119
Gemma 4 E4B
3.7%
120
DeepSeek V3
3.6%
121
Nova Pro
3.4%
122
GPT-4o
3.3%
123
GPT-4 Turbo
3.3%
124
Claude 3 Opus
3.1%

FAQ

What does AA-HLE measure?

A display-only Artificial Analysis Humanity's Last Exam score.

Which model scores highest on AA-HLE?

Claude Opus 4.8 by Anthropic currently leads with a score of 45.7% on AA-HLE.

How many models are evaluated on AA-HLE?

124 AI models have been evaluated on AA-HLE on BenchLM.

Compare Top Models on AA-HLE

Claude Opus 4.8 vs Gemini 3.1 ProGemini 3.1 Pro vs GPT-5.5GPT-5.5 vs GPT-5.4GPT-5.4 vs Gemini 3.5 Flash
Last updated: June 2, 2026 · BenchLM version AA-HLE 2026

The AI models change fast. We track them for you.

For engineers, researchers, and the plain curious — a weekly brief on new models, ranking shifts, and pricing changes.

Free. No spam. Unsubscribe anytime.