Skip to content

Benchmark

GDM-MRCRv2: 8 needle 256k

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

Models measured
45
Scores recorded
128
Sources
1
IQ weight
1.85

Score distribution

39.4%
Mean
39.8%
Median
21%
Std dev
10.4%
10th percentile
69.1%
90th percentile
1.2%
Min
1.2%89%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.85 for chat models, measured over 45 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Gemini 3.7 Flashhigh· Context Arena data89%
2GPT-5.6 Solmax· Context Arena data83.5%
3Gemini 3.7 Flashmedium· Context Arena data82.1%
4Gemini 3.7 Flashlow· Context Arena data79.5%
5GPT 5.5high· Context Arena data75.3%
6Gemini 3.6 Flashhigh· Context Arena data73.9%
7GPT 5.5xhigh· Context Arena data73.2%
8Claude Sonnet 4.6low· Context Arena data72.7%
9GPT 5.5medium· Context Arena data71.9%
10Claude Opus 4.6medium· Context Arena data71.8%
11Qwen3.8 Maxlow· Context Arena data71.2%
12Qwen3.8 Maxmedium· Context Arena data70.6%
13Claude Sonnet 4.6high· Context Arena data69.3%
14Claude Opus 4.6high· Context Arena data69.1%
15Claude Sonnet 4.6medium· Context Arena data68.5%
16Claude Opus 4.6low· Context Arena data68.5%
17GPT-5.6 Terramax· Context Arena data67.9%
18Qwen3.8 Maxxhigh· Context Arena data67.7%
19GPT 5.5low· Context Arena data67%
20Gemini 3.6 Flashmedium· Context Arena data66.4%
21Gemini 3.6 Flashlow· Context Arena data65.7%
22Claude Opus 5max· Context Arena data65.6%
23Claude Opus 4.8max· Context Arena data61.8%
24GLM 5.3 Flashlow· Context Arena data60.1%
25Muse Spark 1.2high· Context Arena data59%
26GLM 5.3max· Context Arena data57.8%
27GLM 5.3high· Context Arena data55.8%
28Muse Spark 1.2medium· Context Arena data55.8%
29GLM 5.3 Flashmax· Context Arena data55.6%
30Gemini 3.5 Flashhigh· Context Arena data54.6%
31Muse Spark 1.2xhigh· Context Arena data52.8%
32GPT 5.4medium· Context Arena data52.7%
33Muse Spark 1.1minimal· Context Arena data52.7%
34GLM 5.3low· Context Arena data52.5%
35Muse Spark 1.2low· Context Arena data52.2%
36Claude Sonnet 5max· Context Arena data52.2%
37Muse Spark 1.2minimal· Context Arena data52.2%
38Muse Spark 1.1medium· Context Arena data51.3%
39GPT 5.4high· Context Arena data51.2%
40Muse Spark 1.1low· Context Arena data49.1%
41GLM 5.3 Flashhigh· Context Arena data48.7%
42Muse Spark 1.1high· Context Arena data48%
43GPT 5.4xhigh· Context Arena data47.8%
44Grok 4.6xhigh· Context Arena data47.7%
45Gemini 3.1 Pro Previewhigh· Context Arena data47.4%
46Grok 4.6medium· Context Arena data46.8%
47Muse Spark 1.1xhigh· Context Arena data46.7%
48GLM 5.2max· Context Arena data46.5%
49Claude Sonnet 4.6· Context Arena data46.1%
50Grok 4.6high· Context Arena data45.8%

Showing 1–50 of 128 results.