Skip to content

Benchmark

GDM-MRCRv2: 8 needle

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
1.21

Score distribution

49.7%
Mean
50.1%
Median
20.2%
Std dev
24%
10th percentile
73.7%
90th percentile
12.3%
Min
12.3%94%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.21 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
1Qwen 3.8 27Bmedium· Context Arena data94%
2Qwen 3.8 27Blow· Context Arena data92.8%
3Qwen 3.8 27Bxhigh· Context Arena data90.2%
4Qwen3.6 Max Previewenabled· Context Arena data88.1%
5GPT-5.6 Solmax· Context Arena data85.5%
6Qwen3.6 35B A3Benabled· Context Arena data83.5%
7Gemini 3.7 Flashhigh· Context Arena data83%
8Qwen3.6 27Benabled· Context Arena data82.2%
9GPT 5.5xhigh· Context Arena data79.8%
10Gemini 3.7 Flashmedium· Context Arena data79.6%
11GPT 5.5high· Context Arena data79%
12GPT 5.5medium· Context Arena data78.6%
13Claude Opus 5max· Context Arena data77.3%
14GPT-5.6 Terramax· Context Arena data76.6%
15Qwen3.5 397B A17Benabled· Context Arena data75.5%
16GPT 5.5low· Context Arena data75%
17Gemini 3.7 Flashlow· Context Arena data74.5%
18Ling 3.0 Flashenabled· Context Arena data74.3%
19Qwen3.8 Maxxhigh· Context Arena data73.7%
20Grok 4.6medium· Context Arena data73.6%
21Grok 4.6xhigh· Context Arena data73.5%
22Gemini 3.6 Flashhigh· Context Arena data73.3%
23Qwen3.5 27Benabled· Context Arena data73.3%
24Claude Opus 4.6medium· Context Arena data73.1%
25Claude Opus 4.8max· Context Arena data72.6%
26Claude Opus 4.6low· Context Arena data72.4%
27Claude Opus 4.6high· Context Arena data72.3%
28Qwen3.8 Maxmedium· Context Arena data72.1%
29Qwen3.8 Maxlow· Context Arena data71.6%
30Grok 4.6high· Context Arena data71.5%
31Qwen3.6 Max Preview· Context Arena data70.6%
32Claude Sonnet 4.6high· Context Arena data70.5%
33GLM 5.3high· Context Arena data70.4%
34Claude Sonnet 4.6low· Context Arena data70.4%
35Gemini 3.6 Flashmedium· Context Arena data69.9%
36Ring 2.6 1Txhigh· Context Arena data69.8%
37Ring 2.6 1Thigh· Context Arena data69.7%
38Claude Sonnet 4.6medium· Context Arena data69.6%
39Qwen3.5 122B-A10Benabled· Context Arena data68.5%
40GLM 5.3max· Context Arena data68.4%
41Muse Spark 1.2high· Context Arena data67.7%
42GPT 5.4xhigh· Context Arena data67.7%
43Muse Spark 1.2medium· Context Arena data67.6%
44Qwen3.5 35B A3Benabled· Context Arena data67.5%
45Muse Spark 1.2low· Context Arena data67.2%
46Grok 4.6low· Context Arena data66.9%
47Hy3 (Deep Infra)high· Context Arena data66.8%
48GPT 5.4high· Context Arena data66.2%
49Hy3 (Deep Infra)low· Context Arena data66.1%
50Muse Spark 1.2xhigh· Context Arena data65.1%

Showing 1–50 of 177 results.