Skip to main content
BenchLM
Advertise

Benchmark profile

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

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

Data verified

Benchmark score on AA-HLE — July 29, 2026

BenchLM mirrors the published score view for AA-HLE. Claude Fable 5 leads the public snapshot at 53.3% , followed by Claude Opus 5 (52.6%) and GPT-5.6 Sol (47.2%). BenchLM does not use these results to rank models overall.

1Closed

Claude Fable 5

Anthropic

claude-fable-5

53.3%
Overall 82.75Context 1M+
2Closed

Claude Opus 5

Anthropic

claude-opus-5

52.6%
Overall 82.81
3Closed

GPT-5.6 Sol

OpenAI

gpt-5-6-sol

47.2%
Overall 81.39Context 1M
163 modelsKnowledgeCurrentDisplay onlyUpdated July 29, 2026

Benchmark score table (163 models)

Score
1
Claude Fable 5Anthropic · Closed
53.3%
2
Claude Opus 5Anthropic · Closed
52.6%
3
GPT-5.6 SolOpenAI · Closed
47.2%
4
Claude Opus 4.8Anthropic · Closed
45.7%
5
Muse Spark 1.1Meta · Closed
45.1%
6
Gemini 3.1 ProGoogle · Closed
44.7%
7
Kimi K3Moonshot AI · Closed
44.3%
8
GPT-5.5OpenAI · Closed
44.3%
9
GPT-5.6 TerraOpenAI · Closed
41.8%
10
GPT-5.4OpenAI · Closed
41.6%
11
Gemini 3.5 FlashGoogle · Closed
41.0%
12
Grok 4.5xAI · Closed
40.3%
13
GLM-5.2Z.AI · Open weight
40.1%
14
GPT-5.3 CodexOpenAI · Closed
39.9%
15
Muse SparkMeta · Closed
39.9%
16
GPT-5.3-Codex-SparkOpenAI · Closed
39.9%
17
Claude Sonnet 5Anthropic · Closed
39.6%
18
Claude Opus 4.7 (Adaptive)Anthropic · Closed
39.6%
19
Gemini 3.6 FlashGoogle · Closed
38.3%
20
Qwen3.7 MaxAlibaba · Closed
38.1%
21
GPT-5.6 LunaOpenAI · Closed
37.2%
22
Gemini 3 ProGoogle · Closed
37.2%
23
MiniMax M3MiniMax · Open weight
37.1%
24
Claude Opus 4.6 (Adaptive)Anthropic · Closed
36.7%
25
DeepSeek V4 Pro (Max)DeepSeek · Open weight
35.9%
26
Kimi K2.6Moonshot AI · Open weight
35.9%
27
GPT-5.2OpenAI · Closed
35.4%
28
Grok 4.3xAI · Closed
35.0%
29
MiMo-V2.5-ProXiaomi · Closed
33.8%
30
DeepSeek V4 Pro (High)DeepSeek · Open weight
33.5%
31
GPT-5.2-CodexOpenAI · Closed
33.5%
32
Qwen3.7 PlusAlibaba · Closed
33.4%
33
Kimi K2.7 CodeMoonshot AI · Open weight
32.8%
34
DeepSeek V4 Flash (Max)DeepSeek · Open weight
32.1%
35
Hy3 PreviewTencent · Open weight
31.6%
36
Hy3Tencent · Open weight
31.6%
37
Claude Opus 4.7Anthropic · Closed
31.2%
38
InklingThinking Machines Lab · Open weight
29.7%
39
Kimi K2.5Moonshot AI · Open weight
29.4%
40
Kimi K2.5 (Reasoning)Moonshot AI · Closed
29.4%
41
Qwen 3.6 Max (preview)Alibaba · Closed
28.9%
42
Claude Opus 4.5 ThinkingAnthropic · Closed
28.4%
43
MiMo-V2-ProXiaomi · Closed
28.3%
44
MiniMax M2.7MiniMax · Open weight
28.1%
45
GLM-5.1Z.AI · Open weight
28.0%
46
DeepSeek V4 Flash (High)DeepSeek · Open weight
27.8%
47
Qwen3.5 397BAlibaba · Open weight
27.3%
48
Qwen3.5 397B (Reasoning)Alibaba · Open weight
27.3%
49
GLM-5Z.AI · Open weight
27.2%
50
GPT-5.4 miniOpenAI · Closed
26.6%
51
Nemotron 3 UltraNVIDIA · Open weight
26.6%
52
GPT-5.4 nanoOpenAI · Closed
26.5%
53
GPT-5.1OpenAI · Closed
26.5%
54
GPT-5 (high)OpenAI · Closed
26.5%
55
Qwen3.6 PlusAlibaba · Closed
25.7%
56
GLM-5-TurboZ.AI · Closed
25.4%
57
GLM-4.7Z.AI · Open weight
25.1%
58
Grok 4xAI · Closed
23.9%
59
GPT-5 (medium)OpenAI · Closed
23.5%
60
Qwen3.5-122B-A10BAlibaba · Open weight
23.4%
61
GPT-5.1-Codex-MaxOpenAI · Closed
23.4%
62
GPT-5.1-CodexOpenAI · Closed
23.4%
63
Gemma 4 31BGoogle · Open weight
22.7%
64
Step 3.5 FlashStepFun · Open weight
22.6%
65
Qwen3.5-27BAlibaba · Open weight
22.2%
66
Qwen3.6-27BAlibaba · Open weight
21.6%
67
Gemini 2.5 ProGoogle · Closed
21.1%
68
Qwen3.6-35B-A3BAlibaba · Open weight
20.2%
69
o3OpenAI · Closed
20.0%
70
Step 3.7 FlashStepFun · Open weight
19.9%
71
MiMo-V2-OmniXiaomi · Closed
19.9%
72
Qwen3.5-35B-A3BAlibaba · Open weight
19.7%
73
GPT-5 miniOpenAI · Closed
19.7%
74
Nemotron 3 Super 120B A12BNVIDIA · Open weight
19.2%
75
MiniMax M2.5MiniMax · Closed
19.1%
76
Claude Opus 4.6Anthropic · Closed
18.6%
77
GPT-OSS 120BOpenAI · Open weight
18.5%
78
Gemma 4 26B A4BGoogle · Open weight
18.3%
79
Grok 4.1 Fast (Reasoning)xAI · Closed
17.6%
80
Gemini 3.5 Flash-LiteGoogle · Closed
17.5%
81
Grok 4 Fast (Reasoning)xAI · Closed
17.0%
82
Gemini 3.1 Flash-LiteGoogle · Closed
16.2%
83
GLM-5V-TurboZ.AI · Closed
15.8%
84
Mercury 2Inception · Closed
15.5%
85
DeepSeek-R1DeepSeek · Open weight
14.9%
86
Gemma 4 12BGoogle · Open weight
14.8%
87
Trinity-Large-PreviewArcee AI · Open weight
14.7%
88
Trinity-Large-ThinkingArcee AI · Open weight
14.7%
89
Gemini 3 FlashGoogle · Closed
14.1%
90
Claude Sonnet 4.6Anthropic · Closed
13.2%
91
K-ExaoneLG AI Research · Closed
13.1%
92
DeepSeek V3.1 (Reasoning)DeepSeek · Open weight
13.0%
93
Claude Opus 4.5Anthropic · Closed
12.9%
94
Mistral Medium 3.5 128BMistral · Open weight
12.8%
95
Claude 4.1 Opus ThinkingAnthropic · Closed
11.9%
96
Command A+Cohere · Open weight
11.4%
97
Qwen3 MaxAlibaba · Closed
11.1%
98
DeepSeek V3.2DeepSeek · Open weight
10.5%
99
Nemotron 3 Nano 30BNVIDIA · Open weight
10.2%
100
Sarvam 105BSarvam · Open weight
10.1%
101
GPT-OSS 20BOpenAI · Open weight
9.8%
102
Mistral Small 4Mistral · Open weight
9.5%
103
Mistral Small 4 (Reasoning)Mistral · Open weight
9.5%
104
o3-miniOpenAI · Closed
8.7%
105
GPT-5 nanoOpenAI · Closed
8.2%
106
MiniMax M1 80kMiniMax · Closed
8.2%
107
Nemotron Ultra 253BNVIDIA · Open weight
8.1%
108
MiMo-V2-FlashXiaomi · Open weight
8.0%
109
o1OpenAI · Closed
7.7%
110
Grok Code Fast 1xAI · Closed
7.5%
111
GLM-4.7-FlashZ.AI · Open weight
7.1%
112
Kimi K2Moonshot AI · Closed
7.0%
113
Sarvam 30BSarvam · Open weight
7.0%
114
LFM2.5-8B-A1BLiquidAI · Open weight
6.9%
115
GLM-4.5-AirZ.AI · Closed
6.8%
116
LFM2.5-1.2B-InstructLiquidAI · Closed
6.8%
117
Granite-4.0-H-350MIBM · Open weight
6.4%
118
DeepSeek V3.1DeepSeek · Open weight
6.3%
119
Ling 2.6 FlashInclusionAI · Open weight
6.2%
120
LFM2.5-1.2B-ThinkingLiquidAI · Closed
6.1%
121
Exaone 4.0 1.2BLG AI Research · Open weight
5.8%
122
Granite-4.0-350MIBM · Open weight
5.7%
123
DeepSeek R1 Distill Qwen 32BDeepSeek · Open weight
5.5%
124
Nemotron 3 Nano Omni 30B A3BNVIDIA · Open weight
5.3%
125
Ministral 3 3B (Reasoning)Mistral · Open weight
5.3%
126
Ministral 3 3BMistral · Open weight
5.3%
127
GLM-4.6Z.AI · Open weight
5.2%
128
Gemini 2.5 FlashGoogle · Closed
5.1%
129
LFM2.5-VL-1.6B-ExtractLiquidAI · Open weight
5.1%
130
Granite-4.0-1BIBM · Open weight
5.1%
131
Grok 4.1 FastxAI · Closed
5.0%
132
Granite-4.0-H-1BIBM · Open weight
5.0%
133
Exaone 4.0 32BLG AI Research · Open weight
4.9%
134
Gemini 1.5 ProGoogle · Closed
4.9%
135
Llama 4 MaverickMeta · Open weight
4.8%
136
Gemma 4 E2BGoogle · Open weight
4.8%
137
Gemma 3 27BGoogle · Open weight
4.7%
138
GPT-4.1OpenAI · Closed
4.6%
139
GPT-4.1 miniOpenAI · Closed
4.6%
140
Gemini 1.0 ProGoogle · Closed
4.6%
141
Ministral 3 14B (Reasoning)Mistral · Open weight
4.6%
142
Ministral 3 14BMistral · Open weight
4.6%
143
LFM2-24B-A2BLiquidAI · Closed
4.4%
144
Llama 4 ScoutMeta · Open weight
4.3%
145
Mistral Medium 3Mistral · Closed
4.3%
146
Ministral 3 8B (Reasoning)Mistral · Open weight
4.3%
147
Ministral 3 8BMistral · Open weight
4.3%
148
Llama 3.1 405BMeta · Open weight
4.2%
149
Mistral Large 3Mistral · Closed
4.1%
150
Phi-4Microsoft · Open weight
4.1%
151
Claude 4 SonnetAnthropic · Closed
4.0%
152
GPT-4o miniOpenAI · Closed
4.0%
153
Mistral Large 2Mistral · Closed
4.0%
154
GPT-4.1 nanoOpenAI · Closed
3.9%
155
Claude 3 HaikuAnthropic · Closed
3.9%
156
Qwen2.5 Coder 32B InstructAlibaba · Open weight
3.8%
157
Solar Pro 2Upstage · Closed
3.8%
158
Gemma 4 E4BGoogle · Open weight
3.7%
159
DeepSeek V3DeepSeek · Open weight
3.6%
160
Nova ProAmazon · Closed
3.4%
161
GPT-4oOpenAI · Closed
3.3%
162
GPT-4 TurboOpenAI · Closed
3.3%
163
Claude 3 OpusAnthropic · Closed
3.1%

The published AA-HLE snapshot places Claude Fable 5 first at 53.3%. The third row is 6.1 points behind. The broader top-10 range is 11.7 points, so the table still separates the published systems.

163 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.

FAQ

What does AA-HLE measure?

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

Which model scores highest on AA-HLE?

Claude Fable 5 by Anthropic currently leads with a score of 53.3% on AA-HLE.

How many models are evaluated on AA-HLE?

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

Compare Top Models on AA-HLE

Claude Fable 5 vs Claude Opus 5Claude Opus 5 vs GPT-5.6 SolGPT-5.6 Sol vs Claude Opus 4.8Claude Opus 4.8 vs Muse Spark 1.1
Last updated: July 29, 2026 · BenchLM version AA-HLE 2026

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 2,000+ readers.

One email each week. Unsubscribe anytime.