Skip to content

Benchmark

GDM-MRCRv2: 8 needle 128k

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
1.80

Score distribution

40.7%
Mean
39.5%
Median
23%
Std dev
13.2%
10th percentile
72.6%
90th percentile
0.51%
Min
0.51%92.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.80 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
1GPT-5.6 Solmax· Context Arena data92.4%
2Qwen3.8 Maxxhigh· Context Arena data92.3%
3Claude Opus 5max· Context Arena data91.3%
4Gemini 3.7 Flashhigh· Context Arena data88.6%
5Qwen3.8 Maxmedium· Context Arena data87.3%
6GPT 5.5xhigh· Context Arena data85.8%
7GPT 5.5medium· Context Arena data83%
8Gemini 3.7 Flashmedium· Context Arena data82.1%
9Qwen 3.8 27Blow· Context Arena data81.8%
10Qwen3.8 Maxlow· Context Arena data81.5%
11GPT-5.6 Terramax· Context Arena data80.2%
12GPT 5.5high· Context Arena data79.3%
13Qwen 3.8 27Bmedium· Context Arena data78.9%
14Gemini 3.7 Flashlow· Context Arena data78.3%
15Gemini 3.6 Flashhigh· Context Arena data77.3%
16Claude Opus 4.8max· Context Arena data75.2%
17Gemini 3.6 Flashmedium· Context Arena data74.5%
18Qwen 3.8 27Bxhigh· Context Arena data74.3%
19Claude Sonnet 4.6low· Context Arena data72.6%
20Claude Opus 4.6medium· Context Arena data72.5%
21Claude Sonnet 4.6high· Context Arena data72.4%
22Claude Sonnet 4.6medium· Context Arena data71.7%
23Qwen3.6 Max Previewenabled· Context Arena data71.6%
24GLM 5.3high· Context Arena data69.3%
25Claude Opus 4.6low· Context Arena data68.8%
26Claude Opus 4.6high· Context Arena data68.2%
27GPT 5.4medium· Context Arena data67.8%
28GPT 5.5low· Context Arena data67%
29GLM 5.3max· Context Arena data65.4%
30GLM 5.3low· Context Arena data64.5%
31GPT 5.4xhigh· Context Arena data63.1%
32GPT 5.4high· Context Arena data63%
33GLM 5.3 Flashhigh· Context Arena data62.9%
34DeepSeek V4 Flash 0731max· Context Arena data62.9%
35DeepSeek V4 Pro 0813max· Context Arena data60.4%
36Muse Spark 1.2low· Context Arena data60.3%
37Muse Spark 1.2medium· Context Arena data59.4%
38Qwen3.7 Flashenabled· Context Arena data59.3%
39Qwen3.6 35B A3Benabled· Context Arena data59.3%
40Qwen3.6 Max Preview· Context Arena data59.2%
41GLM 5.3 Flashmax· Context Arena data58.9%
42GPT-5.6 Lunamax· Context Arena data58.4%
43Gemini 3.1 Pro Previewhigh· Context Arena data57.6%
44Gemini 3.5 Flashhigh· Context Arena data57.5%
45DeepSeek V4 Pro 0813high· Context Arena data57.5%
46Gemini 3.6 Flashlow· Context Arena data57.3%
47Qwen3.6 27Benabled· Context Arena data56.6%
48Qwen3.6 Plusenabled· Context Arena data56.6%
49Muse Spark 1.2high· Context Arena data55.6%
50Gemini 3.5 Flash Litehigh· Context Arena data55.5%

Showing 1–50 of 177 results.