Skip to main content
BenchLM
Advertise

Benchmark profile

Artificial Analysis Long Context Reasoning (AA-LCR)

A display-only Artificial Analysis long-context reasoning evaluation.

Data verified

Benchmark score on AA-LCR — July 23, 2026

BenchLM mirrors the published score view for AA-LCR. GPT-5.2-Codex leads the public snapshot at 75.7% , followed by GPT-5 (high) (75.6%) and GPT-5.1 (75.0%). BenchLM does not use these results to rank models overall.

1Closed

GPT-5.2-Codex

OpenAI

gpt-5-2-codex

75.7%
Overall 59.1Context 400K
2Closed

GPT-5 (high)

OpenAI

gpt-5-high

75.6%
Overall 58.61Context 128K
3Closed

GPT-5.1

OpenAI

gpt-5-1

75.0%
Overall 53.65Context 200K
155 modelsReasoningCurrentDisplay onlyUpdated July 23, 2026

Benchmark score table (155 models)

Score
1
GPT-5.2-CodexOpenAI · Closed
75.7%
2
GPT-5 (high)OpenAI · Closed
75.6%
3
GPT-5.1OpenAI · Closed
75.0%
4
Kimi K3Moonshot AI · Closed
74.7%
5
GPT-5.5OpenAI · Closed
74.3%
6
GPT-5.6 TerraOpenAI · Closed
74.0%
7
GPT-5.6 LunaOpenAI · Closed
74.0%
8
GPT-5.4OpenAI · Closed
74.0%
9
MiniMax M3MiniMax · Open weight
74.0%
10
GPT-5.3 CodexOpenAI · Closed
74.0%
11
Claude Opus 4.5 ThinkingAnthropic · Closed
74.0%
12
GPT-5.3-Codex-SparkOpenAI · Closed
74.0%
13
GPT-5.6 SolOpenAI · Closed
73.7%
14
MiMo-V2.5-ProXiaomi · Closed
73.3%
15
GPT-5 (medium)OpenAI · Closed
72.8%
16
Gemini 3.1 ProGoogle · Closed
72.7%
17
GPT-5.2OpenAI · Closed
72.7%
18
GLM-5.2Z.AI · Open weight
71.3%
19
Claude Sonnet 5Anthropic · Closed
70.7%
20
Gemini 3 ProGoogle · Closed
70.7%
21
Claude Opus 4.6 (Adaptive)Anthropic · Closed
70.7%
22
Claude Opus 4.7 (Adaptive)Anthropic · Closed
70.3%
23
Claude Fable 5Anthropic · Closed
70.0%
24
Kimi K2.6Moonshot AI · Open weight
69.7%
25
Qwen3.6 PlusAlibaba · Closed
69.7%
26
Gemini 3.6 FlashGoogle · Closed
69.7%
27
Qwen 3.6 Max (preview)Alibaba · Closed
69.7%
28
Muse SparkMeta · Closed
69.7%
29
Gemini 3.5 FlashGoogle · Closed
69.3%
30
GPT-5.4 miniOpenAI · Closed
69.3%
31
o3OpenAI · Closed
69.3%
32
Qwen3.7 MaxAlibaba · Closed
69.0%
33
Qwen3.6-27BAlibaba · Open weight
68.7%
34
MiniMax M2.7MiniMax · Open weight
68.7%
35
Grok 4xAI · Closed
68.0%
36
Grok 4.1 Fast (Reasoning)xAI · Closed
68.0%
37
GPT-5 miniOpenAI · Closed
68.0%
38
Claude Opus 4.8Anthropic · Closed
67.7%
39
Grok 4.5xAI · Closed
67.7%
40
Qwen3.5-27BAlibaba · Open weight
67.3%
41
GPT-5.1-Codex-MaxOpenAI · Closed
67.3%
42
GPT-5.1-CodexOpenAI · Closed
67.3%
43
Nemotron 3 UltraNVIDIA · Open weight
67.0%
44
Claude Opus 4.7Anthropic · Closed
67.0%
45
Qwen3.5-122B-A10BAlibaba · Open weight
66.7%
46
Hy3 PreviewTencent · Open weight
66.7%
47
MiMo-V2-OmniXiaomi · Closed
66.7%
48
Hy3Tencent · Open weight
66.7%
49
DeepSeek V4 Pro (Max)DeepSeek · Open weight
66.3%
50
Claude 4.1 Opus ThinkingAnthropic · Closed
66.3%
51
Kimi K2.7 CodeMoonshot AI · Open weight
66.3%
52
GPT-5.4 nanoOpenAI · Closed
66.0%
53
Gemini 2.5 ProGoogle · Closed
66.0%
54
MiniMax M2.5MiniMax · Closed
66.0%
55
Qwen3.5 397BAlibaba · Open weight
65.7%
56
Qwen3.5 397B (Reasoning)Alibaba · Open weight
65.7%
57
Claude Opus 4.5Anthropic · Closed
65.3%
58
Kimi K2.5Moonshot AI · Open weight
65.3%
59
Kimi K2.5 (Reasoning)Moonshot AI · Closed
65.3%
60
Gemini 3.1 Flash-LiteGoogle · Closed
65.3%
61
Qwen3.7 PlusAlibaba · Closed
65.0%
62
DeepSeek V4 Pro (High)DeepSeek · Open weight
65.0%
63
Grok 4 Fast (Reasoning)xAI · Closed
64.7%
64
Grok 4.3xAI · Closed
64.3%
65
GLM-4.7Z.AI · Open weight
64.0%
66
Qwen3.6-35B-A3BAlibaba · Open weight
63.7%
67
Step 3.7 FlashStepFun · Open weight
63.7%
68
Muse Spark 1.1Meta · Closed
63.3%
69
GLM-5Z.AI · Open weight
63.3%
70
InklingThinking Machines Lab · Open weight
63.3%
71
DeepSeek V4 Flash (Max)DeepSeek · Open weight
63.0%
72
DeepSeek V4 Flash (High)DeepSeek · Open weight
62.7%
73
Qwen3.5-35B-A3BAlibaba · Open weight
62.7%
74
GLM-5.1Z.AI · Open weight
62.3%
75
Gemini 3.5 Flash-LiteGoogle · Closed
62.0%
76
Gemma 4 31BGoogle · Open weight
62.0%
77
Mistral Medium 3.5 128BMistral · Open weight
61.0%
78
GPT-4.1OpenAI · Closed
61.0%
79
GLM-5V-TurboZ.AI · Closed
61.0%
80
GLM-5-TurboZ.AI · Closed
60.7%
81
MiMo-V2-ProXiaomi · Closed
60.7%
82
Nemotron 3 Super 120B A12BNVIDIA · Open weight
60.0%
83
o1OpenAI · Closed
59.3%
84
Claude Opus 4.6Anthropic · Closed
58.3%
85
Claude Sonnet 4.6Anthropic · Closed
57.7%
86
Gemma 4 26B A4BGoogle · Open weight
55.7%
87
K-ExaoneLG AI Research · Closed
55.7%
88
Gemma 4 12BGoogle · Open weight
55.3%
89
DeepSeek-R1DeepSeek · Open weight
54.7%
90
Step 3.5 FlashStepFun · Open weight
54.3%
91
MiniMax M1 80kMiniMax · Closed
54.3%
92
DeepSeek V3.1 (Reasoning)DeepSeek · Open weight
53.3%
93
Kimi K2Moonshot AI · Closed
51.0%
94
GPT-OSS 120BOpenAI · Open weight
50.7%
95
Grok Code Fast 1xAI · Closed
48.3%
96
Gemini 3 FlashGoogle · Closed
48.0%
97
Qwen3 MaxAlibaba · Closed
46.7%
98
Llama 4 MaverickMeta · Open weight
46.0%
99
Command A+Cohere · Open weight
46.0%
100
Gemini 2.5 FlashGoogle · Closed
45.9%
101
DeepSeek V3.1DeepSeek · Open weight
45.0%
102
Mistral Small 4Mistral · Open weight
44.7%
103
Mistral Small 4 (Reasoning)Mistral · Open weight
44.7%
104
Claude 4 SonnetAnthropic · Closed
44.3%
105
GLM-4.5-AirZ.AI · Closed
43.7%
106
GPT-4.1 miniOpenAI · Closed
42.3%
107
GPT-5 nanoOpenAI · Closed
41.7%
108
DeepSeek V3.2DeepSeek · Open weight
39.0%
109
Mercury 2Inception · Closed
36.3%
110
Nemotron 3 Nano Omni 30B A3BNVIDIA · Open weight
35.7%
111
GLM-4.7-FlashZ.AI · Open weight
35.0%
112
Mistral Large 3Mistral · Closed
34.7%
113
Nemotron 3 Nano 30BNVIDIA · Open weight
33.7%
114
Trinity-Large-PreviewArcee AI · Open weight
33.0%
115
Trinity-Large-ThinkingArcee AI · Open weight
33.0%
116
MiMo-V2-FlashXiaomi · Open weight
31.3%
117
GPT-OSS 20BOpenAI · Open weight
30.7%
118
Gemma 4 E4BGoogle · Open weight
30.7%
119
DeepSeek V3DeepSeek · Open weight
29.0%
120
Mistral Medium 3Mistral · Closed
28.0%
121
GLM-4.6Z.AI · Open weight
26.3%
122
Llama 4 ScoutMeta · Open weight
25.8%
123
Ling 2.6 FlashInclusionAI · Open weight
25.0%
124
Llama 3.1 405BMeta · Open weight
24.3%
125
Ministral 3 8B (Reasoning)Mistral · Open weight
24.0%
126
Ministral 3 8BMistral · Open weight
24.0%
127
Grok 4.1 FastxAI · Closed
22.0%
128
Ministral 3 14B (Reasoning)Mistral · Open weight
22.0%
129
Ministral 3 14BMistral · Open weight
22.0%
130
Claude 3 HaikuAnthropic · Closed
21.0%
131
Nova ProAmazon · Closed
19.0%
132
GPT-4.1 nanoOpenAI · Closed
17.0%
133
Gemma 4 E2BGoogle · Open weight
15.0%
134
Ministral 3 3B (Reasoning)Mistral · Open weight
11.7%
135
Ministral 3 3BMistral · Open weight
11.7%
136
DeepSeek R1 Distill Qwen 32BDeepSeek · Open weight
9.7%
137
Exaone 4.0 32BLG AI Research · Open weight
8.0%
138
Nemotron Ultra 253BNVIDIA · Open weight
7.3%
139
Granite-4.0-H-1BIBM · Open weight
6.3%
140
Gemma 3 27BGoogle · Open weight
5.7%
141
Mistral Large 2Mistral · Closed
5.3%
142
Granite-4.0-1BIBM · Open weight
4.0%
143
LFM2.5-8B-A1BLiquidAI · Open weight
0.0%
144
GPT-4oOpenAI · Closed
0.0%
145
Phi-4Microsoft · Open weight
0.0%
146
LFM2.5-VL-1.6B-ExtractLiquidAI · Open weight
0.0%
147
Sarvam 105BSarvam · Open weight
0.0%
148
Sarvam 30BSarvam · Open weight
0.0%
149
Granite-4.0-350MIBM · Open weight
0.0%
150
Granite-4.0-H-350MIBM · Open weight
0.0%
151
Exaone 4.0 1.2BLG AI Research · Open weight
0.0%
152
Solar Pro 2Upstage · Closed
0.0%
153
LFM2-24B-A2BLiquidAI · Closed
0.0%
154
LFM2.5-1.2B-ThinkingLiquidAI · Closed
0.0%
155
LFM2.5-1.2B-InstructLiquidAI · Closed
0.0%

The published AA-LCR snapshot places GPT-5.2-Codex first at 75.7%. The third row is 0.7 points behind. The broader top-10 range is 1.7 points, so many of the published results sit in a relatively narrow band.

155 models have been evaluated on AA-LCR. The benchmark falls in the Reasoning category. This category carries a 17% weight in BenchLM.ai's overall scoring system. AA-LCR is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About AA-LCR

Year

2026

Tasks

Long-context reasoning tasks

Format

Accuracy

Difficulty

Long-context reasoning

BenchLM stores AA-LCR as a display-only row when OpenRouter or Artificial Analysis publishes the exact long-context reasoning card value.

Artificial Analysis model benchmarks

BenchLM freshness & provenance

Version

AA-LCR 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-LCR measure?

A display-only Artificial Analysis long-context reasoning evaluation.

Which model scores highest on AA-LCR?

GPT-5.2-Codex by OpenAI currently leads with a score of 75.7% on AA-LCR.

How many models are evaluated on AA-LCR?

155 AI models have been evaluated on AA-LCR on BenchLM.

Compare Top Models on AA-LCR

GPT-5.2-Codex vs GPT-5 (high)GPT-5 (high) vs GPT-5.1GPT-5.1 vs Kimi K3Kimi K3 vs GPT-5.5
Last updated: July 23, 2026 · BenchLM version AA-LCR 2026

Choose a model with this week’s evidence

Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.

One email each week. Unsubscribe anytime.