Artificial Analysis Long Context Reasoning (AA-LCR)
Data verified
An Artificial Analysis long-context reasoning evaluation.
Top models on AA-LCR — November 4, 2026
As of November 4, 2026, Kimi K3 leads the AA-LCR leaderboard with 88.7% , followed by Step 5 Preview (88.3%) and Kimi K3.1 Pro (86.3%).
1Open weights
Kimi K3
Moonshot AI
88.7%
Overall 69.34Context 1.05M
2Open weights
Step 5 Preview
StepFun
88.3%
Overall 56.15Context 1M
3Open weights
Kimi K3.1 Pro
Moonshot AI
86.3%
Overall 74.61Context 1.05M
234 modelsReasoning15% of category scoreCurrentUpdated November 4, 2026
Leaderboard (234 models)
234 of 234 rows
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Kimi K3Moonshot AI | 88.7% | Not reported | Open |
| 2 | Step 5 PreviewStepFun | 88.3% | Not reported | Open |
| 3 | Kimi K3.1 ProMoonshot AI | 86.3% | Not reported | Open |
| 4 | MiMo-V2.6-ProXiaomi | 86.3% | Not reported | Open |
| 5 | Claude Fable 5.1Anthropic | 85.3% | Not reported | Closed |
| 6 | Claude Opus 5.5Anthropic | 84.7% | Not reported | Closed |
| 7 | GPT-5.5OpenAI | 84.3% | Not reported | Closed |
| 8 | GPT-5.6 SolOpenAI | 84.0% | Not reported | Closed |
| 9 | DeepSeek V4.1 Flash 1017DeepSeek | 84.0% | Not reported | Open |
| 10 | GPT-5.6 LunaOpenAI | 83.7% | Not reported | Closed |
| 11 | GPT-6 SolOpenAI | 83.7% | Not reported | Closed |
| 12 | Muse Wonder FullMeta | 83.7% | Not reported | Closed |
| 13 | GPT-5.3 CodexOpenAI | 83.3% | Not reported | Closed |
| 14 | Muse Glimmer 30BMeta | 83.3% | Not reported | Open |
| 15 | GPT-6 LunaOpenAI | 83.3% | Not reported | Closed |
| 16 | GPT-5.3-Codex-SparkOpenAI | 83.3% | Not reported | Closed |
| 17 | GPT-5.6 TerraOpenAI | 83.0% | Not reported | Closed |
| 18 | MiniMax M3MiniMax | 83.0% | Not reported | Open |
| 19 | GPT-6.1 SolOpenAI | 83.0% | Not reported | Closed |
| 20 | Ling 3.1 FlashInclusionAI | 83.0% | Not reported | Open |
| 21 | Muse Spark 1.3Meta | 83.0% | Not reported | Closed |
| 22 | Claude Sonnet 5.5Anthropic | 82.7% | Not reported | Closed |
| 23 | Claude Haiku 5.5Anthropic | 82.7% | Not reported | Closed |
| 24 | GPT-5.2OpenAI | 82.7% | Not reported | Closed |
| 25 | GPT-6.1 TerraOpenAI | 82.7% | Not reported | Closed |
| 26 | Claude Fable 5Anthropic | 82.3% | Not reported | Closed |
| 27 | GPT-5.2-CodexOpenAI | 82.3% | Not reported | Closed |
| 28 | Claude Sonnet 5Anthropic | 82.0% | Not reported | Closed |
| 29 | Gemini 3.1 ProGoogle | 82.0% | Not reported | Closed |
| 30 | GPT-5.4OpenAI | 82.0% | Not reported | Closed |
| 31 | Qwen3.8-27BAlibaba | 82.0% | Not reported | Open |
| 32 | Gemini 3.7 FlashGoogle | 81.7% | Not reported | Closed |
| 33 | Gemini 3.8 FlashGoogle | 81.3% | Not reported | Closed |
| 34 | Mistral Large 4Mistral | 81.3% | Not reported | Open |
| 35 | Kimi K2.6Moonshot AI | 81.0% | Not reported | Open |
| 36 | Sage-A1.5InternScience | 81.0% | Not reported | Open |
| 37 | GPT-6 AstraOpenAI | 80.7% | Not reported | Closed |
| 38 | Muse WonderMeta | 80.7% | Not reported | Closed |
| 39 | Qwen 3.6 Max (preview)Alibaba | 80.7% | Not reported | Closed |
| 40 | DeepSeek V4 Pro 0813DeepSeek | 80.3% | Not reported | Open |
| 41 | Grok 4.6xAI | 80.3% | Not reported | Closed |
| 42 | Qwen3.8 Max PreviewAlibaba | 80.3% | Not reported | Closed |
| 43 | GPT-5.1OpenAI | 80.0% | Not reported | Closed |
| 44 | Gemini 3.6 FlashGoogle | 80.0% | Not reported | Closed |
| 45 | Gemini 4 ArgonGoogle | 79.7% | Not reported | Closed |
| 46 | MiniMax M3.1 FlashMiniMax | 79.7% | Not reported | Closed |
| 47 | MiMo-V2.5-ProXiaomi | 79.7% | Not reported | Closed |
| 48 | GLM-5.3Z.AI | 79.7% | Not reported | Open |
| 49 | Qwen3.8-Flash-NextAlibaba | 79.7% | Not reported | Open |
| 50 | DeepSeek V4 Flash 0731DeepSeek | 79.7% | Not reported | Open |
| 51 | Beam 1.0 501B-A23B FP8Reflection AI | 79.4% | Not reported | Open |
| 52 | Claude Opus 5Anthropic | 79.3% | Not reported | Closed |
| 53 | Kimi K2.7 CodeMoonshot AI | 79.3% | Not reported | Open |
| 54 | Gemini 4 CarbonGoogle | 79.3% | Not reported | Closed |
| 55 | Apodex 1.1Apodex | 79.3% | Not reported | Closed |
| 56 | Grok 4.5xAI | 79.3% | Not reported | Closed |
| 57 | Apodex 1.1 MiniApodex | 79.3% | Not reported | Open |
| 58 | Qwen3.7 MaxAlibaba | 79.0% | Not reported | Closed |
| 59 | Hy3Tencent | 79.0% | Not reported | Open |
| 60 | Muse Spark 1.2Meta | 79.0% | Not reported | Closed |
| 61 | Claude Opus 4.7 (Adaptive)Anthropic | 78.7% | Not reported | Closed |
| 62 | Beam 1.0 501B-A23B NVFP4Reflection AI | 78.7% | Not reported | Open |
| 63 | Qwen3.6 PlusAlibaba | 78.3% | Not reported | Closed |
| 64 | GLM-5.2Z.AI | 78.3% | Not reported | Open |
| 65 | MiniMax M2.7MiniMax | 78.3% | Not reported | Open |
| 66 | Ling 3.0 Flash VLInclusionAI | 78.3% | Not reported | Open |
| 67 | Muse Spark OpenMeta | 78.3% | Not reported | Open |
| 68 | Beam 1.0 501B-A23BReflection AI | 78.2% | Not reported | Open |
| 69 | GPT-5 (high)OpenAI | 78.2% | Not reported | Closed |
| 70 | Muse SparkMeta | 78.0% | Not reported | Closed |
| 71 | Kimi K2.5 (Reasoning)Moonshot AI | 78.0% | Not reported | Closed |
| 72 | Kimi K2.5Moonshot AI | 78.0% | Not reported | Open |
| 73 | Claude Opus 4.6 (Adaptive)Anthropic | 78.0% | Not reported | Closed |
| 74 | Claude Opus 4.8Anthropic | 77.7% | Not reported | Closed |
| 75 | Muse Spark 1.1Meta | 77.7% | Not reported | Closed |
| 76 | Qwen3.5-27BAlibaba | 77.7% | Not reported | Open |
| 77 | InklingThinking Machines Lab | 77.3% | Not reported | Open |
| 78 | Qwen3.6-27BAlibaba | 77.3% | Not reported | Open |
| 79 | Qwen3.5 397BAlibaba | 77.3% | Not reported | Open |
| 80 | Claude Opus 4.5 ThinkingAnthropic | 77.3% | Not reported | Closed |
| 81 | Qwen3.5 397B (Reasoning)Alibaba | 77.3% | Not reported | Open |
| 82 | GPT-5.4 miniOpenAI | 77.0% | Not reported | Closed |
| 83 | Ternary Bonsai 2 27BPrism ML | 77.0% | Not reported | Open |
| 84 | GPT-5.4 nanoOpenAI | 76.7% | Not reported | Closed |
| 85 | Grok 4.7xAI | 76.7% | Not reported | Closed |
| 86 | Qwen3.5-122B-A10BAlibaba | 76.3% | Not reported | Open |
| 87 | Quasar 438BMultiverse Computing | 76.3% | Not reported | Closed |
| 88 | Gemini 3 ProGoogle | 76.0% | Not reported | Closed |
| 89 | GPT-5 (medium)OpenAI | 76.0% | Not reported | Closed |
| 90 | Claude 4.1 Opus ThinkingAnthropic | 76.0% | Not reported | Closed |
| 91 | Gemini 3.5 Flash-LiteGoogle | 76.0% | Not reported | Closed |
| 92 | Inkling-SmallThinking Machines Lab | 75.7% | Not reported | Open |
| 93 | GLM-5Z.AI | 75.7% | Not reported | Open |
| 94 | Claude Opus 4.7Anthropic | 75.7% | Not reported | Closed |
| 95 | MiMo-V2-OmniXiaomi | 75.0% | Not reported | Closed |
| 96 | o3OpenAI | 74.7% | Not reported | Closed |
| 97 | MiMo-V2.6-FlashXiaomi | 74.3% | Not reported | Open |
| 98 | Grok 4.1 Fast (Reasoning)xAI | 74.0% | Not reported | Closed |
| 99 | GLM-5.1Z.AI | 73.7% | Not reported | Open |
| 100 | Step 3.7 FlashStepFun | 73.7% | Not reported | Open |
| 101 | Grok 4 Fast (Reasoning)xAI | 73.7% | Not reported | Closed |
| 102 | MiniMax M2.5MiniMax | 73.3% | Not reported | Closed |
| 103 | Qwen3.7 PlusAlibaba | 73.0% | Not reported | Closed |
| 104 | Ling 3.0 FlashInclusionAI | 73.0% | Not reported | Open |
| 105 | GPT-5 miniOpenAI | 72.3% | Not reported | Closed |
| 106 | Qwen3.5-35B-A3BAlibaba | 72.0% | Not reported | Open |
| 107 | Qwen3.6-35B-A3BAlibaba | 71.7% | Not reported | Open |
| 108 | GLM-5-TurboZ.AI | 71.7% | Not reported | Closed |
| 109 | GLM-4.7Z.AI | 71.0% | Not reported | Open |
| 110 | Solar Pro 4Upstage | 71.0% | Not reported | Closed |
| 111 | Claude Opus 4.5Anthropic | 70.7% | Not reported | Closed |
| 112 | Ling 3.0 Flash SanteInclusionAI | 70.5% | Not reported | Open |
| 113 | GLM-5V-TurboZ.AI | 70.3% | Not reported | Closed |
| 114 | Gemma 4 31BGoogle | 69.7% | Not reported | Open |
| 115 | Nemotron 3.5 Bolt 74B-A9BNVIDIA | 69.7% | Not reported | Open |
| 116 | Gemini 3.5 FlashGoogle | 69.3% | Not reported | Closed |
| 117 | GPT-5.1-CodexOpenAI | 69.3% | Not reported | Closed |
| 118 | Claude 4 Opus ThinkingAnthropic | 69.3% | Not reported | Closed |
| 119 | GPT-5.1-Codex-MaxOpenAI | 69.3% | Not reported | Closed |
| 120 | Mistral Medium 3.5 128BMistral | 69.3% | Not reported | Open |
| 121 | Gemini 2.5 ProGoogle | 69.0% | Not reported | Closed |
| 122 | Nemotron 3.5 Super VL 120B A12BNVIDIA | 68.6% | Not reported | Open |
| 123 | Claude Sonnet 4.6Anthropic | 68.3% | Not reported | Closed |
| 124 | GPT-4.1OpenAI | 68.3% | Not reported | Closed |
| 125 | MiMo-V2-ProXiaomi | 68.3% | Not reported | Closed |
| 126 | Grok 4xAI | 68.0% | Not reported | Closed |
| 127 | Mercury 2.5Inception | 68.0% | Not reported | Closed |
| 128 | Nemotron 3.5 Super VL 120B A12B NVFP4NVIDIA | 67.7% | Not reported | Open |
| 129 | Nemotron 3.5 Bolt 74B-A9B BF16NVIDIA | 67.4% | Not reported | Open |
| 130 | Spark-X2.6-4BSparkLLM | 67.3% | Not reported | Open |
| 131 | Claude Opus 4.6Anthropic | 67.0% | Not reported | Closed |
| 132 | Nemotron 3 UltraNVIDIA | 67.0% | Not reported | Open |
| 133 | DeepSeek V4 Pro (High)DeepSeek | 67.0% | Not reported | Open |
| 134 | Hy3 PreviewTencent | 66.7% | Not reported | Open |
| 135 | A.X K2SK Telecom | 66.0% | Not reported | Open |
| 136 | Gemma 4 26B A4BGoogle | 65.7% | Not reported | Open |
| 137 | o1OpenAI | 65.0% | Not reported | Closed |
| 138 | Grok 4.3xAI | 64.3% | Not reported | Closed |
| 139 | Gemma 4 12BGoogle | 63.7% | Not reported | Open |
| 140 | Qwen3.5-4BAlibaba | 63.0% | Not reported | Open |
| 141 | Step 3.5 FlashStepFun | 63.0% | Not reported | Open |
| 142 | Solar Open 2Upstage | 62.3% | Not reported | Open |
| 143 | K-ExaoneLG AI Research | 61.3% | Not reported | Closed |
| 144 | o4-miniOpenAI | 61.0% | Not reported | Closed |
| 145 | Ling 3.0 TinyInclusionAI | 60.3% | Not reported | Open |
| 146 | MiniCPM5-2BOpenBMB | 59.0% | Not reported | Open |
| 147 | MiniMax M1 80kMiniMax | 57.7% | Not reported | Closed |
| 148 | DeepSeek V3.1 (Reasoning)DeepSeek | 56.7% | Not reported | Open |
| 149 | K-EXAONE 2.0LG AI Research | 56.2% | Not reported | Open |
| 150 | DeepSeek-R1DeepSeek | 55.7% | Not reported | Open |
| 151 | Gemini 3 FlashGoogle | 55.3% | Not reported | Closed |
| 152 | Kimi K2Moonshot AI | 53.0% | Not reported | Closed |
| 153 | Grok Code Fast 1xAI | 53.0% | Not reported | Closed |
| 154 | LFM3-1.7BLiquidAI | 52.8% | Not reported | Open |
| 155 | Command A+Cohere | 52.7% | Not reported | Open |
| 156 | GPT-OSS 120BOpenAI | 52.0% | Not reported | Open |
| 157 | Claude 3.7 SonnetAnthropic | 51.7% | Not reported | Closed |
| 158 | Llama 4 MaverickMeta | 50.0% | Not reported | Open |
| 159 | Qwen3 MaxAlibaba | 50.0% | Not reported | Closed |
| 160 | Gemini 2.5 FlashGoogle | 49.9% | Not reported | Closed |
| 161 | Mistral Small 4Mistral | 49.7% | Not reported | Open |
| 162 | GPT-4oOpenAI | 49.3% | Not reported | Closed |
| 163 | Nemotron 3.5 Lightning 30B A3B NVFP4NVIDIA | 49.2% | Not reported | Open |
| 164 | Granite 4.2 30BIBM | 49.0% | Not reported | Open |
| 165 | DeepSeek V3.1DeepSeek | 47.0% | Not reported | Open |
| 166 | GLM-4.5-AirZ.AI | 46.7% | Not reported | Closed |
| 167 | DeepSeek V3.2DeepSeek | 45.7% | Not reported | Open |
| 168 | Granite 4.2 8BIBM | 45.0% | Not reported | Open |
| 169 | GPT-5 nanoOpenAI | 45.0% | Not reported | Closed |
| 170 | Claude 4 SonnetAnthropic | 44.0% | Not reported | Closed |
| 171 | GPT-4.1 miniOpenAI | 44.0% | Not reported | Closed |
| 172 | Mercury 2Inception | 43.7% | Not reported | Closed |
| 173 | GLM-4.7-FlashZ.AI | 41.7% | Not reported | Open |
| 174 | DeepSeek V3 0324DeepSeek | 40.7% | Not reported | Open |
| 175 | Nemotron 3 Nano Omni 30B A3BNVIDIA | 39.7% | Not reported | Open |
| 176 | Celeris-1Celeris | 38.0% | Not reported | Closed |
| 177 | Trinity-Large-ThinkingArcee AI | 38.0% | Not reported | Open |
| 178 | Trinity-Large-PreviewArcee AI | 38.0% | Not reported | Open |
| 179 | North Mini CodeCohere | 37.3% | Not reported | Open |
| 180 | MiMo-V2-FlashXiaomi | 36.3% | Not reported | Open |
| 181 | Mistral Large 3Mistral | 36.0% | Not reported | Closed |
| 182 | GPT-OSS 20BOpenAI | 34.7% | Not reported | Open |
| 183 | Solar Pro 3Upstage | 32.3% | Not reported | Closed |
| 184 | Gemma 4 E4BGoogle | 32.0% | Not reported | Open |
| 185 | Gemini 2.5 Flash-LiteGoogle | 32.0% | Not reported | Closed |
| 186 | Mistral Medium 3Mistral | 31.3% | Not reported | Closed |
| 187 | Grok 4.1 FastxAI | 31.3% | Not reported | Closed |
| 188 | Gemini 2.0 FlashGoogle | 31.3% | Not reported | Closed |
| 189 | Ling 2.6 FlashInclusionAI | 31.3% | Not reported | Open |
| 190 | DeepSeek V3DeepSeek | 29.3% | Not reported | Open |
| 191 | Mistral Small 4 (Reasoning)Mistral | 28.3% | Not reported | Open |
| 192 | Llama 4 ScoutMeta | 27.7% | Not reported | Open |
| 193 | Claude 3 HaikuAnthropic | 27.7% | Not reported | Closed |
| 194 | Claude 3.5 HaikuAnthropic | 27.3% | Not reported | Closed |
| 195 | QwQ-32BAlibaba | 26.7% | Not reported | Open |
| 196 | GLM-4.6Z.AI | 26.3% | Not reported | Open |
| 197 | Ministral 3 14B (Reasoning)Mistral | 26.3% | Not reported | Open |
| 198 | Ministral 3 14BMistral | 26.3% | Not reported | Open |
| 199 | Ministral 3 8B (Reasoning)Mistral | 25.7% | Not reported | Open |
| 200 | Ministral 3 8BMistral | 25.7% | Not reported | Open |
| 201 | Llama 3.1 405BMeta | 25.3% | Not reported | Open |
| 202 | Granite 4.2 3BIBM | 24.3% | Not reported | Open |
| 203 | Mistral Small 3.1Mistral | 22.3% | Not reported | Open |
| 204 | Nova ProAmazon | 21.0% | Not reported | Closed |
| 205 | GPT-4.1 nanoOpenAI | 20.3% | Not reported | Closed |
| 206 | Devstral SmallMistral | 18.7% | Not reported | Open |
| 207 | Ministral 3 3B (Reasoning)Mistral | 17.0% | Not reported | Open |
| 208 | Ministral 3 3BMistral | 17.0% | Not reported | Open |
| 209 | Gemma 4 E2BGoogle | 16.3% | Not reported | Open |
| 210 | Llama 3.3 70BMeta | 15.7% | Not reported | Open |
| 211 | Ultravox v0.6 Llama 3.3 70BFixie AI | 15.7% | Not reported | Open |
| 212 | Nemotron 3 Nano 30BNVIDIA | 10.7% | Not reported | Open |
| 213 | DeepSeek R1 Distill Qwen 32BDeepSeek | 8.7% | Not reported | Open |
| 214 | Gemma 3 27BGoogle | 7.3% | Not reported | Open |
| 215 | Granite-4.0-H-1BIBM | 7.0% | Not reported | Open |
| 216 | LFM2.5-2.6BLiquidAI | 5.7% | Not reported | Open |
| 217 | Llama 3.1 70BMeta | 4.7% | Not reported | Open |
| 218 | Phi-4Microsoft | 0.0% | Not reported | Open |
| 219 | Phi-4-reasoningMicrosoft | 0.0% | Not reported | Open |
| 220 | Magistral MediumMistral | 0.0% | Not reported | Closed |
| 221 | Qwen3-Omni-30B-A3B-InstructAlibaba | 0.0% | Not reported | Open |
| 222 | Magistral SmallMistral | 0.0% | Not reported | Open |
| 223 | Solar Pro 2Upstage | 0.0% | Not reported | Closed |
| 224 | Exaone 4.0 1.2BLG AI Research | 0.0% | Not reported | Open |
| 225 | Sarvam 105BSarvam | 0.0% | Not reported | Open |
| 226 | LFM2.5-8B-A1BLiquidAI | 0.0% | Not reported | Open |
| 227 | Granite-4.0-H-350MIBM | 0.0% | Not reported | Open |
| 228 | Granite-4.0-350MIBM | 0.0% | Not reported | Open |
| 229 | Sarvam 30BSarvam | 0.0% | Not reported | Open |
| 230 | LFM2.5-VL-1.6B-ExtractLiquidAI | 0.0% | Not reported | Open |
| 231 | Qwen3-Omni-30B-A3B-ThinkingAlibaba | 0.0% | Not reported | Open |
| 232 | LFM2-24B-A2BLiquidAI | 0.0% | Not reported | Closed |
| 233 | LFM2.5-1.2B-ThinkingLiquidAI | 0.0% | Not reported | Closed |
| 234 | LFM2.5-1.2B-InstructLiquidAI | 0.0% | Not reported | Closed |
According to BenchLM.ai, Kimi K3 leads the AA-LCR benchmark with a score of 88.7%, followed by Step 5 Preview (88.3%) and Kimi K3.1 Pro (86.3%). The top models are clustered within 2.4 points, suggesting this benchmark is nearing saturation for frontier models.
234 models have been evaluated on AA-LCR. The benchmark falls in the Reasoning category. The Reasoning leaderboard ranks models by a weighted category score, and AA-LCR contributes 15% of it. BenchAlign v5.8 also uses it in the overall ranking.
About AA-LCR
Year
2026
Tasks
Long-context reasoning tasks
Format
Accuracy
Difficulty
Long-context reasoning
BenchLM stores AA-LCR when OpenRouter or Artificial Analysis publishes the exact long-context reasoning card value.
Freshness and provenance
Version
AA-LCR 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
Current
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.
Questions
What does AA-LCR measure?
An Artificial Analysis long-context reasoning evaluation.
Which model scores highest on AA-LCR?
Kimi K3 by Moonshot AI currently leads with a score of 88.7% on AA-LCR.
How many models are evaluated on AA-LCR?
234 AI models have published results on AA-LCR in the BenchLM catalog.
Last updated: November 4, 2026 · BenchLM version AA-LCR 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 issueJoin 6,100+ subscribers.
One email each week. Unsubscribe anytime.