Why this benchmark is useful
Long-context research breaks when models lose thread across distant evidence. AA-LCR is one of the few public suites that grades synthesis across long inputs instead of retrieval trivia alone.
Artificial Analysis Long Context Reasoning
AA-LCR
Long-context synthesis and reasoning: 100 open-answer questions requiring models to integrate evidence across long inputs.
What this does not measure
- Retrieval from external databases — the context is in-prompt only.
- Multimodal long documents — AA-LCR is text-only in the current AA harness.
- Infinite context — the eval uses a fixed long-context window, not arbitrary document length.
Coverage map
- Task family
- Long context
- Format
- 100 open-answer questions with long context passages; equality-checker LLM grading, pass@1.
- Scoring
- pass@1 accuracy with 3 repeats.
- Maintainer
- Artificial Analysis
How to read the scores
Read pass@1 with three repeats as a conservative headline. Compare only rows from the same AA harness version and long-context window.
What it does not cover
- Retrieval from external databases — the context is in-prompt only.
- Multimodal long documents — AA-LCR is text-only in the current AA harness.
- Infinite context — the eval uses a fixed long-context window, not arbitrary document length.
Scores
184 rowsEvidence ledger
184184 rows
88.67%best score
184source-checked
1sources
2026-09-28to 2026-06-17
Where the rows land
0255075100
Normalized to this benchmark's axis (0–100). Open the table for raw units.
Newest receipts
- Kimi K388.67% ·
- MiMo-V2.6-Pro86.33% ·
- Claude Fable 5.185.33% ·
- Claude Opus 5.584.67% ·
- GPT-5.5 (high)84.33% ·
Scores use this benchmark's own unit and axis, not a universal quality score.
| # | Model | Release date | Score | Provenance | Trust |
|---|---|---|---|---|---|
| 1 | Kimi K3Moonshot | 2026-07-16 | 88.67% | 2026-09-28 | Source-checked |
| 2 | MiMo-V2.6-ProXiaomi | 2026-09-22 | 86.33% | 2026-09-28 | Source-checked |
| 3 | Claude Fable 5.1Anthropic | 2026-09-01 | 85.33% | 2026-09-28 | Source-checked |
| 4 | Claude Opus 5.5Anthropic | 2026-09-22 | 84.67% | 2026-09-28 | Source-checked |
| 5 | GPT-5.5 (high)OpenAI | 2026-04-23 | 84.33% | 2026-09-28 | Source-checked |
| 6 | DeepSeek V4.1 FlashDeepSeek | 2026-09-10 | 84% | 2026-09-28 | Source-checked |
| 7 | GPT-5.6 SolOpenAI | 2026-07-09 | 84% | 2026-09-28 | Source-checked |
| 8 | GPT-5.6 LunaOpenAI | 2026-07-09 | 83.67% | 2026-09-28 | Source-checked |
| 9 | GPT-6 SolOpenAI | 2026-09-22 | 83.67% | 2026-09-28 | Source-checked |
| 10 | GPT-5.3 Codex (xhigh)OpenAI | 2026-02-05 | 83.33% | 2026-09-28 | Source-checked |
| 11 | GPT-6 LunaOpenAI | 2026-09-22 | 83.33% | 2026-09-28 | Source-checked |
| 12 | Muse GlimmerMeta | 2026-08-10 | 83.33% | 2026-09-28 | Source-checked |
| 13 | GPT-5.5 (medium)OpenAI | 2026-04-23 | 83% | 2026-09-28 | Source-checked |
| 14 | GPT-5.6 TerraOpenAI | 2026-07-09 | 83% | 2026-09-28 | Source-checked |
| 15 | MiniMax M3MiniMax | 2026-05-31 | 83% | 2026-09-28 | Source-checked |
| 16 | Muse Spark 1.3Meta | 2026-09-02 | 83% | 2026-09-28 | Source-checked |
| 17 | GPT-5.2OpenAI | 2025-12-11 | 82.67% | 2026-09-28 | Source-checked |
| 18 | Claude Fable 5Anthropic | 2026-06-09 | 82.33% | 2026-09-28 | Source-checked |
| 19 | GPT-5.2 Codex (xhigh)OpenAI | 2025-12-11 | 82.33% | 2026-09-28 | Source-checked |
| 20 | Claude Opus 5 (Adaptive Reasoning, Medium Effort)Anthropic | 2026-07-24 | 82% | 2026-09-28 | Source-checked |
| 21 | Claude Sonnet 5Anthropic | 2026-06-30 | 82% | 2026-09-28 | Source-checked |
| 22 | Gemini 3.1 Pro PreviewGoogle | 2026-02-19 | 82% | 2026-09-28 | Source-checked |
| 23 | GPT 5.4OpenAI | 2026-03-05 | 82% | 2026-09-28 | Source-checked |
| 24 | Qwen3.8 27BAlibaba | 2026-08-14 | 82% | 2026-09-28 | Source-checked |
| 25 | Gemini 3.7 FlashGoogle | 2026-08-13 | 81.67% | 2026-09-28 | Source-checked |
| 26 | Nex-N2-ProOther | 2026-06-02 | 81.67% | 2026-09-28 | Source-checked |
| 27 | Claude Opus 5 (Adaptive Reasoning, Low Effort)Anthropic | 2026-07-24 | 81.33% | 2026-09-28 | Source-checked |
| 28 | DeepSeek V4 Flash Vision (Reasoning, Max Effort)DeepSeek | 2026-08-21 | 81.33% | 2026-09-28 | Source-checked |
| 29 | Gemini 3.8 FlashGoogle | 2026-09-02 | 81.33% | 2026-09-28 | Source-checked |
| 30 | GPT-5.5 (low)OpenAI | 2026-04-23 | 81% | 2026-09-28 | Source-checked |
| 31 | Kimi K2.6Moonshot | 2026-04-20 | 81% | 2026-09-28 | Source-checked |
| 32 | GPT-6 AstraOpenAI | 2026-09-03 | 80.67% | 2026-09-28 | Source-checked |
| 33 | Qwen3.6 Max PreviewAlibaba | 2026-04-20 | 80.67% | 2026-09-28 | Source-checked |
| 34 | Claude Opus 5 (Adaptive Reasoning, Xhigh Effort)Anthropic | 2026-07-24 | 80.33% | 2026-09-28 | Source-checked |
| 35 | DeepSeek V4 Pro 0813 (Reasoning, Max Effort)DeepSeek | 2026-08-13 | 80.33% | 2026-09-28 | Source-checked |
| 36 | Grok 4.6xAI | 2026-08-12 | 80.33% | 2026-09-28 | Source-checked |
| 37 | Qwen3.8 2.4T A95BAlibaba | 2026-08-12 | 80.33% | 2026-09-28 | Source-checked |
| 38 | Qwen3.8 MaxAlibaba | 2026-08-03 | 80.33% | 2026-09-28 | Source-checked |
| 39 | Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)Anthropic | 2026-02-17 | 80% | 2026-09-28 | Source-checked |
| 40 | Gemini 3.6 Flash (high)Google | 2026-07-21 | 80% | 2026-09-28 | Source-checked |
| 41 | GLM-5.3-FlashZhipu | 2026-08-26 | 80% | 2026-09-28 | Source-checked |
| 42 | GPT-5.1OpenAI | 2025-11-13 | 80% | 2026-09-28 | Source-checked |
| 43 | K2 Horizon 375B A23BOther | 2026-09-03 | 80% | 2026-09-28 | Source-checked |
| 44 | DeepSeek V4 Flash 0731 (Reasoning, Max Effort)DeepSeek | 2026-07-31 | 79.67% | 2026-09-28 | Source-checked |
| 45 | GLM-5.3Zhipu | 2026-08-14 | 79.67% | 2026-09-28 | Source-checked |
| 46 | MiMo-V2.5-ProXiaomi | 2026-04-22 | 79.67% | 2026-09-28 | Source-checked |
| 47 | Qwen3.8-Flash-NextAlibaba | 2026-08-26 | 79.67% | 2026-09-28 | Source-checked |
| 48 | Apodex 1.1Other | 2026-08-30 | 79.33% | 2026-09-28 | Source-checked |
| 49 | Claude Opus 5 (Adaptive Reasoning, Max Effort)Anthropic | 2026-07-24 | 79.33% | 2026-09-28 | Source-checked |
| 50 | Grok 4.5xAI | 2026-07-08 | 79.33% | 2026-09-28 | Source-checked |
| 51 | Kimi K2.7 CodeMoonshot | 2026-06-12 | 79.33% | 2026-09-28 | Source-checked |
| 52 | Kimi K3 (low)Moonshot | 2026-07-16 | 79.33% | 2026-09-28 | Source-checked |
| 53 | Nemotron 3 Ultra 550B A55B (Reasoning)NVIDIA | 2026-06-04 | 79.33% | 2026-09-28 | Source-checked |
| 54 | Claude Opus 5 (Adaptive Reasoning, High Effort)Anthropic | 2026-07-24 | 79% | 2026-09-28 | Source-checked |
| 55 | Hy3Other | 2026-07-06 | 79% | 2026-09-28 | Source-checked |
| 56 | Muse Spark 1.2Meta | 2026-08-05 | 79% | 2026-09-28 | Source-checked |
| 57 | Qwen3.7 MaxAlibaba | 2026-05-19 | 79% | 2026-09-28 | Source-checked |
| 58 | Claude Opus 4.7Anthropic | 2026-04-16 | 78.67% | 2026-09-28 | Source-checked |
| 59 | GLM-5.2Zhipu | 2026-06-13 | 78.33% | 2026-09-28 | Source-checked |
| 60 | MiniMax M2.7MiniMax | 2026-04-15 | 78.33% | 2026-09-28 | Source-checked |
| 61 | Qwen3.6 PlusAlibaba | 2026-04-02 | 78.33% | 2026-09-28 | Source-checked |
| 62 | GPT-5 (high)OpenAI | 2025-08-07 | 78.2% | 2026-09-28 | Source-checked |
| 63 | Claude Opus 4.6 (Adaptive Reasoning, Max Effort)Anthropic | 2026-02-05 | 78% | 2026-09-28 | Source-checked |
| 64 | Gemini 3 Flash Preview (Reasoning)Google | 2025-12-17 | 78% | 2026-09-28 | Source-checked |
| 65 | Kimi K2.5Moonshot | 2026-01-27 | 78% | 2026-09-28 | Source-checked |
| 66 | Muse SparkOther | 2026-01-01 | 78% | 2026-09-28 | Source-checked |
| 67 | Muse Spark 1.1Meta | 2026-07-09 | 77.67% | 2026-09-28 | Source-checked |
| 68 | Qwen3.5 27B (Reasoning)Alibaba | 2026-02-24 | 77.67% | 2026-09-28 | Source-checked |
| 69 | Claude Opus 4.5 (Reasoning)Anthropic | 2025-11-24 | 77.33% | 2026-09-28 | Source-checked |
| 70 | Inkling (xhigh)Other | 2026-07-15 | 77.33% | 2026-09-28 | Source-checked |
| 71 | Qwen3.5 397B A17B (Reasoning)Alibaba | 2026-02-16 | 77.33% | 2026-09-28 | Source-checked |
| 72 | Qwen3.6 27B (Reasoning)Alibaba | 2026-04-22 | 77.33% | 2026-09-28 | Source-checked |
| 73 | GPT-5.4 MiniOpenAI | 2026-03-17 | 77% | 2026-09-28 | Source-checked |
| 74 | GPT-5.4 (low)OpenAI | 2026-03-05 | 76.67% | 2026-09-28 | Source-checked |
| 75 | GPT-5.4 NanoOpenAI | 2026-03-17 | 76.67% | 2026-09-28 | Source-checked |
| 76 | Grok 4.7xAI | 2026-09-21 | 76.67% | 2026-09-28 | Source-checked |
| 77 | Qwen3.5 122B A10B (Reasoning)Alibaba | 2026-02-24 | 76.33% | 2026-09-28 | Source-checked |
| 78 | Claude 4.1 Opus (Reasoning)Anthropic | 2025-08-05 | 76% | 2026-09-28 | Source-checked |
| 79 | Gemini 3 ProGoogle | 2025-11-18 | 76% | 2026-09-28 | Source-checked |
| 80 | Gemini 3.5 Flash-LiteGoogle | 2026-07-21 | 76% | 2026-09-28 | Source-checked |
| 81 | GPT-5 (medium)OpenAI | 2025-08-07 | 76% | 2026-09-28 | Source-checked |
| 82 | Motif 3Other | 2026-08-12 | 76% | 2026-09-28 | Source-checked |
| 83 | Claude Opus 4.7 (Non-reasoning, High Effort)Anthropic | 2026-04-16 | 75.67% | 2026-09-28 | Source-checked |
| 84 | GLM-5 (Reasoning)Zhipu | 2026-02-11 | 75.67% | 2026-09-28 | Source-checked |
| 85 | Inkling SmallOther | 2026-07-30 | 75.67% | 2026-09-28 | Source-checked |
| 86 | Grok 4.3 (medium)xAI | 2026-04-30 | 75% | 2026-09-28 | Source-checked |
| 87 | MiMo-V2-OmniXiaomi | 2026-03-19 | 75% | 2026-09-28 | Source-checked |
| 88 | Grok Build 0.1 0616xAI | 2026-06-16 | 74.67% | 2026-09-28 | Source-checked |
| 89 | Motif 3 (Beta)Other | 2026-07-21 | 74.67% | 2026-09-28 | Source-checked |
| 90 | OpenAI o3OpenAI | 2025-04-16 | 74.67% | 2026-09-28 | Source-checked |
| 91 | Gemini 3.1 Flash LiteGoogle | 2026-05-07 | 74.33% | 2026-09-28 | Source-checked |
| 92 | Gemini 3.5 Flash (medium)Google | 2026-05-19 | 74.33% | 2026-09-28 | Source-checked |
| 93 | MiMo-V2.6-FlashXiaomi | 2026-09-22 | 74.33% | 2026-09-28 | Source-checked |
| 94 | Qwen3 Max ThinkingAlibaba | 2026-01-26 | 74.33% | 2026-09-28 | Source-checked |
| 95 | Gemini 3 Pro Preview (low)Google | 2025-11-18 | 74% | 2026-09-28 | Source-checked |
| 96 | Grok 4.3 (low)xAI | 2026-04-30 | 74% | 2026-09-28 | Source-checked |
| 97 | Solar Pro 4Other | 2026-08-06 | 74% | 2026-09-28 | Source-checked |
| 98 | GLM-5.1Zhipu | 2026-04-07 | 73.67% | 2026-09-28 | Source-checked |
| 99 | Step 3.7 FlashStepFun | 2026-06-01 | 73.67% | 2026-09-28 | Source-checked |
| 100 | DeepSeek V3.2 (Reasoning)DeepSeek | 2025-12-01 | 73.33% | 2026-09-28 | Source-checked |
| 101 | Gemini 3.5 FlashGoogle | 2026-05-19 | 73.33% | 2026-09-28 | Source-checked |
| 102 | MiMo-V2-Omni-0327Xiaomi | 2026-03-27 | 73.33% | 2026-09-28 | Source-checked |
| 103 | MiniMax M2.5MiniMax | 2026-04-01 | 73.33% | 2026-09-28 | Source-checked |
| 104 | Grok 4.3xAI | 2026-04-30 | 73% | 2026-09-28 | Source-checked |
| 105 | KAT Coder Pro V2Other | 2026-03-27 | 73% | 2026-09-28 | Source-checked |
| 106 | Ling 3.0 FlashOther | 2026-08-04 | 73% | 2026-09-28 | Source-checked |
| 107 | MiMo-V2.5Xiaomi | 2026-04-22 | 73% | 2026-09-28 | Source-checked |
| 108 | Qwen 3.7 PlusAlibaba | 2026-06-02 | 73% | 2026-09-28 | Source-checked |
| 109 | Claude 4.5 Sonnet (Reasoning)Anthropic | 2025-09-29 | 72.33% | 2026-09-28 | Source-checked |
| 110 | Kimi K2 ThinkingMoonshot | 2025-11-06 | 72% | 2026-09-28 | Source-checked |
| 111 | GLM-5-TurboZhipu | 2026-03-15 | 71.67% | 2026-09-28 | Source-checked |
| 112 | GPT-5 Codex (high)OpenAI | 2025-09-23 | 71.67% | 2026-09-28 | Source-checked |
| 113 | Qwen3.6 35B A3B (Reasoning)Alibaba | 2026-04-16 | 71.67% | 2026-09-28 | Source-checked |
| 114 | Solar Open2 250BOther | 2026-08-12 | 71.67% | 2026-09-28 | Source-checked |
| 115 | JT-4.1 Flash 236B A21BOther | 2026-07-09 | 71.33% | 2026-09-28 | Source-checked |
| 116 | MiMo-V2-Flash (Feb 2026)Xiaomi | 2025-12-16 | 71.33% | 2026-09-28 | Source-checked |
| 117 | GLM-4.7 (Reasoning)Zhipu | 2025-12-22 | 71% | 2026-09-28 | Source-checked |
| 118 | Claude Opus 4.5Anthropic | 2025-11-24 | 70.67% | 2026-09-28 | Source-checked |
| 119 | MiMo-V2-Flash (Reasoning)Xiaomi | 2025-12-16 | 70.67% | 2026-09-28 | Source-checked |
| 120 | GLM 5V Turbo (Reasoning)Zhipu | 2026-04-01 | 70.33% | 2026-09-28 | Source-checked |
| 121 | GPT-5.2 (medium)OpenAI | 2025-12-11 | 70.33% | 2026-09-28 | Source-checked |
| 122 | A.X-K2Other | 2026-08-12 | 70% | 2026-09-28 | Source-checked |
| 123 | GPT-5.5 Instant (June 2026)OpenAI | 2026-06-25 | 70% | 2026-09-28 | Source-checked |
| 124 | Ring-2.6-1TOther | 2026-05-08 | 70% | 2026-09-28 | Source-checked |
| 125 | Gemma 4 31B ITGoogle | 2026-03-01 | 69.67% | 2026-09-28 | Source-checked |
| 126 | Kimi K2.6 (Non-reasoning)Other | 2026-04-20 | 69.67% | 2026-09-28 | Source-checked |
| 127 | Claude 4 Opus (Reasoning)Anthropic | 2025-05-22 | 69.33% | 2026-09-28 | Source-checked |
| 128 | Claude Sonnet 4.6 (Non-reasoning, Low Effort)Anthropic | 2026-02-17 | 69.33% | 2026-09-28 | Source-checked |
| 129 | GPT-5.1 Codex (high)OpenAI | 2025-11-13 | 69.33% | 2026-09-28 | Source-checked |
| 130 | Mistral Medium 3.5Mistral | 2026-04-29 | 69.33% | 2026-09-28 | Source-checked |
| 131 | Grok 4.20 ReasoningxAI | 2026-03-05 | 69% | 2026-09-28 | Source-checked |
| 132 | Claude Sonnet 4.6Anthropic | 2026-02-17 | 68.33% | 2026-09-28 | Source-checked |
| 133 | MiMo-V2-ProXiaomi | 2026-03-18 | 68.33% | 2026-09-28 | Source-checked |
| 134 | Grok 4xAI | 2025-07-10 | 68% | 2026-09-28 | Source-checked |
| 135 | Grok 4.20 0309 (Reasoning)xAI | 2026-03-10 | 67.67% | 2026-09-28 | Source-checked |
| 136 | MiniMax-M2.1MiniMax | 2025-12-23 | 67.67% | 2026-09-28 | Source-checked |
| 137 | Claude Opus 4.6Anthropic | 2026-02-05 | 67% | 2026-09-28 | Source-checked |
| 138 | Qwen3.6 27B (Non-reasoning)Alibaba | 2026-04-22 | 66.67% | 2026-09-28 | Source-checked |
| 139 | GPT-5.5 Instant (May 2026)OpenAI | 2026-05-05 | 66.33% | 2026-09-28 | Source-checked |
| 140 | JT-35B-FlashOther | 2026-05-14 | 65.67% | 2026-09-28 | Source-checked |
| 141 | G9v3-39A5BOther | 2026-08-03 | 65.33% | 2026-09-28 | Source-checked |
| 142 | LongCat-2.0Meituan | 2026-06-29 | 65% | 2026-09-28 | Source-checked |
| 143 | OpenAI o1OpenAI | 2024-09-12 | 65% | 2026-09-28 | Source-checked |
| 144 | Hy3-preview (Reasoning)Other | 2026-04-23 | 64.67% | 2026-09-28 | Source-checked |
| 145 | Qwen3.5 397B A17B (Non-reasoning)Alibaba | 2026-02-16 | 64.33% | 2026-09-28 | Source-checked |
| 146 | Qwen3.6 35B A3B (Non-reasoning)Alibaba | 2026-04-16 | 64.33% | 2026-09-28 | Source-checked |
| 147 | GPT-5.5 (Non-reasoning)OpenAI | 2026-04-23 | 64% | 2026-09-28 | Source-checked |
| 148 | Gemma 4 12B (Reasoning)Google | 2026-06-07 | 63.67% | 2026-09-28 | Source-checked |
| 149 | Gemini 3.5 Flash (minimal)Google | 2026-05-19 | 61.33% | 2026-09-28 | Source-checked |
| 150 | Ling 3.0 TinyOther | 2026-08-06 | 60.33% | 2026-09-28 | Source-checked |
| 151 | Nemotron 3.5 LightningNVIDIA | 2026-08-11 | 60.33% | 2026-09-28 | Source-checked |
| 152 | K-EXAONE 2.0 0803Other | 2026-08-12 | 60% | 2026-09-28 | Source-checked |
| 153 | MiniCPM5-2BOpenBMB | 2026-09-07 | 59.33% | 2026-09-28 | Source-checked |
| 1 | Claude Opus 4.8Anthropic | 2026-05-28 | 58.3% | 2026-06-17 | Source-checked |
| 155 | DeepSeek R1DeepSeek | 2025-01-20 | 55.67% | 2026-09-28 | Source-checked |
| 156 | Gemini 3 Flash PreviewGoogle | 2025-12-17 | 55.33% | 2026-09-28 | Source-checked |
| 157 | EXAONE 4.5 33BOther | 2026-04-09 | 54.67% | 2026-09-28 | Source-checked |
| 2 | GPT 5.5OpenAI | 2026-04-23 | 52.1% | 2026-06-17 | Source-checked |
| 159 | Llama 4 MaverickMeta | 2026-04-05 | 50% | 2026-09-28 | Source-checked |
| 160 | Granite 4.2 30BOther | 2026-08-25 | 49% | 2026-09-28 | Source-checked |
| 161 | Granite 4.2 8BOther | 2026-08-25 | 45% | 2026-09-28 | Source-checked |
| 162 | GLM-5 (Non-reasoning)Zhipu | 2026-02-11 | 43.67% | 2026-09-28 | Source-checked |
| 163 | GLM-5.2 (Non-reasoning)Zhipu | 2026-06-16 | 42.33% | 2026-09-28 | Source-checked |
| 164 | Hy3-preview (Non-reasoning)Other | 2026-04-23 | 42% | 2026-09-28 | Source-checked |
| 165 | Ling-2.6-1TOther | 2026-04-23 | 41.67% | 2026-09-28 | Source-checked |
| 166 | MiMo-V2.5-Pro (Non-reasoning)Xiaomi | 2026-04-22 | 41.67% | 2026-09-28 | Source-checked |
| 167 | G9v3-3BOther | 2026-07-23 | 41.33% | 2026-09-28 | Source-checked |
| 168 | Nemotron 3 Nano Omni 30B A3B ReasoningNVIDIA | 2026-04-29 | 39.67% | 2026-09-28 | Source-checked |
| 169 | North Mini CodeCohere | 2026-06-09 | 37.33% | 2026-09-28 | Source-checked |
| 170 | Mistral Large 3Mistral | 2025-12-02 | 36% | 2026-09-28 | Source-checked |
| 171 | Gemma 4 12B (Non-reasoning)Google | 2026-06-10 | 35% | 2026-09-28 | Source-checked |
| 172 | Grok 4.3 (Non-reasoning)xAI | 2026-04-30 | 32.33% | 2026-09-28 | Source-checked |
| 173 | Ling 2.6 FlashOther | 2026-04-21 | 31.33% | 2026-09-28 | Source-checked |
| 174 | Llama 4 ScoutMeta | 2026-04-05 | 27.7% | 2026-09-28 | Source-checked |
| 175 | Granite 4.2 3BOther | 2026-08-25 | 24.33% | 2026-09-28 | Source-checked |
| 176 | Granite 4.1 30BOther | 2026-04-29 | 23.67% | 2026-09-28 | Source-checked |
| 177 | Grok 4.20 0309 v2 (Non-reasoning)xAI | 2026-04-07 | 23.33% | 2026-09-28 | Source-checked |
| 178 | Command ACohere | 2026-03-15 | 21.33% | 2026-09-28 | Source-checked |
| 179 | DiffusionGemma 26B A4BGoogle | 2026-06-10 | 19.67% | 2026-09-28 | Source-checked |
| 180 | JT-MINIOther | 2026-04-15 | 13% | 2026-09-28 | Source-checked |
| 181 | MiniCPM-V 4.6 1.3BOpenBMB | 2026-05-11 | 7% | 2026-09-28 | Source-checked |
| 182 | MiniCPM5-1B (Reasoning)OpenBMB | 2026-06-04 | 6% | 2026-09-28 | Source-checked |
| 183 | MiniCPM5-1B (Non-reasoning)OpenBMB | 2026-05-25 | 5% | 2026-09-28 | Source-checked |
| 184 | LFM2.5-8B-A1BLiquid | 2026-06-08 | 0% | 2026-09-28 | Source-checked |
What it covers
This benchmark sits in the long context family. It uses 100 open-answer questions with long context passages; equality-checker LLM grading, pass@1. The score should travel with its task format, scoring method, source date, and benchmark version.
Where it breaks down
High scores on in-prompt long context do not prove reliable retrieval from external corpora or multimodal documents.
Receipts
Sources and further reading
AA-LCR evaluation· checked 2026-06-22
Questions about this benchmark
Every answer below is assembled from the dated fields on this page. Nothing is written separately for search.
What does AA-LCR measure?
Long-context synthesis and reasoning: 100 open-answer questions requiring models to integrate evidence across long inputs.
What does a high AA-LCR score not prove?
A strong AA-LCR result says nothing about:
- Retrieval from external databases — the context is in-prompt only.
- Multimodal long documents — AA-LCR is text-only in the current AA harness.
- Infinite context — the eval uses a fixed long-context window, not arbitrary document length.
How is AA-LCR scored?
pass@1 accuracy with 3 repeats. Task format: 100 open-answer questions with long context passages; equality-checker LLM grading, pass@1.
Is AA-LCR saturated?
AA-LCR is currently marked Active in the atlas.
Can AA-LCR results be contaminated by training data?
Contamination risk for AA-LCR is graded Low contamination. Treat every row on this page as a public claim with a source and a date, not as a controlled experiment.
The Pack · Editorial newsletter
Know when the rules move.
One short email when a university changes its AI rule for theses, or a research tool changes its pricing or privacy terms. No tracking, no third-party analytics. Unsubscribe in one click.