Skip to content

Benchmark

GDM-MRCRv2: 8 needle 32k

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
1.85

Score distribution

60.7%
Mean
65.5%
Median
27.2%
Std dev
23.8%
10th percentile
94.2%
90th percentile
11.8%
Min
11.8%99.3%
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 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
101Gemini 3.1 Flash Litemedium· Context Arena data56.1%
102MiMo V2.5 Proenabled· Context Arena data54.6%
103Qwen3.6 35B A3B· Context Arena data54.4%
104GPT 5.4 Nanoxhigh· Context Arena data54.1%
105Qwen3.6 Plus· Context Arena data54.1%
106Qwen3.7 Flash· Context Arena data52.3%
107MiniMax M3enabled· Context Arena data52.2%
108Qwen3.6 Flash· Context Arena data52.2%
109GPT-5.4 minihigh· Context Arena data51.8%
110Qwen 3.8 27B· Context Arena data51.5%
111Kimi K2.6· Context Arena data51.4%
112GPT-5.4 minixhigh· Context Arena data50.9%
113Ling 3.0 Flash· Context Arena data49.6%
114GPT-5.4 minimedium· Context Arena data49.5%
115Claude Opus 4.7medium· Context Arena data49.4%
116Claude Sonnet 4.6· Context Arena data49.3%
117Qwen3.6 27B· Context Arena data46.6%
118Claude Opus 4.7xhigh· Context Arena data46.3%
119Claude Opus 4.7low· Context Arena data45.3%
120Gemini 3 Flashlow· Context Arena data43.4%
121Claude Opus 4.7high· Context Arena data42.5%
122Step 3.7 Flashhigh· Context Arena data39.1%
123MiMo V2 Omnienabled· Context Arena data38.1%
124Step 3.7 Flashmedium· Context Arena data37.7%
125Qwen3.5 9Benabled· Context Arena data37.2%
126GPT-5.4 minilow· Context Arena data35.8%
127GPT 5.4 Nanohigh· Context Arena data35.6%
128MiMo V2 Proenabled· Context Arena data35.6%
129MiniMax M2.7enabled· Context Arena data35.5%
130GPT 5.5· Context Arena data35%
131Gemini 3.1 Flash Litelow· Context Arena data34.7%
132MiMo V2.5enabled· Context Arena data34.7%
133Qwen3.5 27B· Context Arena data32.6%
134Nemotron 3 Super 120B A12B· Context Arena data32.3%
135Qwen3.5 35B A3B· Context Arena data31.7%
136Claude Haiku 4.5enabled· Context Arena data31.5%
137Inkling Smallhigh· Context Arena data31.3%
138Inkling Smallmax· Context Arena data31.2%
139DeepSeek V4 Pro 0813· Context Arena data30.7%
140Qwen3.5 122B-A10B· Context Arena data30.4%
141Inkling Smallminimal· Context Arena data29.1%
142Inkling Smallmedium· Context Arena data29%
143Qwen3.5 397B A17B· Context Arena data28.9%
144Gemini 3 Flash· Context Arena data28.4%
145Qwen3.5 Plus· Context Arena data27.4%
146GPT 5.4 Nanomedium· Context Arena data27.2%
147MiMo V2.5· Context Arena data27%
148Step 3.7 Flashlow· Context Arena data26.6%
149DeepSeek V4 Pro· Context Arena data26.5%
150Gemini 3.6 Flashminimal· Context Arena data26.5%

Showing 101–150 of 177 results.