Skip to content

Benchmark

GDM-MRCRv2: 8 needle 16k

Long context benchmarks measure how well models use information in long inputs. Scores are percentages.

Models measured
68
Scores recorded
177
Sources
1
IQ weight
2.62

Score distribution

69.7%
Mean
79.3%
Median
25.6%
Std dev
34.2%
10th percentile
97.5%
90th percentile
13.6%
Min
13.6%100%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.62 for chat models, measured over 68 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
51Gemini 3 Flashhigh· Context Arena data92.6%
52GPT 5.4high· Context Arena data92.6%
53Grok 4.6high· Context Arena data91.5%
54Muse Spark 1.2high· Context Arena data91.5%
55DeepSeek V4 Pro 0813max· Context Arena data91.5%
56DeepSeek V4 Flash 0731high· Context Arena data91.4%
57Qwen3.7 Flashenabled· Context Arena data91.4%
58GLM 5.3low· Context Arena data90.2%
59Kimi K3max· Context Arena data89.9%
60Claude Opus 4.6low· Context Arena data89.3%
61Claude Opus 4.6medium· Context Arena data89.2%
62GPT 5.4low· Context Arena data87.8%
63Muse Spark 1.1medium· Context Arena data87.8%
64Claude Opus 4.7high· Context Arena data86.4%
65Claude Opus 4.7medium· Context Arena data86.2%
66Muse Spark 1.1xhigh· Context Arena data85.5%
67DeepSeek V4 Flash 0731max· Context Arena data85.4%
68Qwen3.6 Max Previewenabled· Context Arena data85.4%
69DeepSeek V4 Flashenabled· Context Arena data84.8%
70Claude Opus 4.7low· Context Arena data84.6%
71Claude Opus 4.6high· Context Arena data84.4%
72Gemini 3.1 Flash Litehigh· Context Arena data84.2%
73GPT 5.4medium· Context Arena data84.2%
74Claude Opus 4.7xhigh· Context Arena data84%
75DeepSeek V4 Flash 0731low· Context Arena data83.9%
76Muse Spark 1.1low· Context Arena data82.9%
77Kimi K3high· Context Arena data82.9%
78GLM 5.3 Flashmax· Context Arena data82.6%
79Ring 2.6 1Txhigh· Context Arena data82.4%
80Claude Sonnet 4.6low· Context Arena data82.2%
81Claude Sonnet 4.6medium· Context Arena data82%
82Claude Sonnet 4.6high· Context Arena data82%
83Muse Spark 1.1high· Context Arena data81.8%
84GLM 5.2max· Context Arena data81.8%
85Gemini 3.5 Flash Litehigh· Context Arena data80.7%
86Kimi K2.6enabled· Context Arena data80.6%
87Gemini 3.1 Flash Litemedium· Context Arena data80.1%
88GLM 5.3 Flashhigh· Context Arena data79.3%
89GLM 5.1enabled· Context Arena data79.3%
90Ling 3.0 Flashenabled· Context Arena data78.6%
91Doubao Seed 2.1 Turboenabled· Context Arena data78.3%
92Qwen3.6 Max Preview· Context Arena data78.1%
93Ring 2.6 1Thigh· Context Arena data77.6%
94Gemini 3.5 Flash Litelow· Context Arena data76.9%
95Kimi K2.5enabled· Context Arena data76.9%
96Muse Spark 1.1minimal· Context Arena data76.7%
97Gemini 3 Flashlow· Context Arena data75.7%
98Gemini 3.5 Flash Litemedium· Context Arena data73.4%
99Kimi K3low· Context Arena data71.9%
100GPT-5.4 minihigh· Context Arena data70.8%

Showing 51–100 of 177 results.