Skip to content

Benchmark

GDM-MRCRv2: 8 needle 64k

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
1.74

Score distribution

50.4%
Mean
49.6%
Median
24.4%
Std dev
20%
10th percentile
84.1%
90th percentile
8.9%
Min
8.9%100%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.74 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
101Hy3 (Deep Infra)high· Context Arena data42.5%
102Ling 3.0 Flash· Context Arena data40.6%
103Qwen3.6 35B A3B· Context Arena data40.5%
104MiMo V2.5 Proenabled· Context Arena data40.4%
105Gemini 3.1 Flash Litehigh· Context Arena data39%
106Grok 4.20enabled· Context Arena data39%
107MiMo V2.5 Pro· Context Arena data38.4%
108GPT 5.5· Context Arena data37.6%
109GPT-5.4 minixhigh· Context Arena data37.3%
110DeepSeek V4 Flash 0731low· Context Arena data36.9%
111Gemini 3.1 Flash Litelow· Context Arena data36.9%
112Kimi K2.5· Context Arena data35.9%
113GPT 5.4 Nanoxhigh· Context Arena data35.6%
114MiniMax M3enabled· Context Arena data35.4%
115Qwen3.5 Flash· Context Arena data35.4%
116DeepSeek V4 Pro 0813· Context Arena data35%
117Kimi K2.5enabled· Context Arena data34.9%
118Qwen3.5 27B· Context Arena data34.8%
119GPT-5.4 minimedium· Context Arena data34.7%
120Gemini 3 Flashlow· Context Arena data34.6%
121MiMo V2 Proenabled· Context Arena data32.4%
122Qwen3.5 397B A17B· Context Arena data32.2%
123Qwen3.5 Plus· Context Arena data32.2%
124GPT-5.4 minihigh· Context Arena data31.8%
125Claude Opus 4.7low· Context Arena data31%
126Qwen3.5 9Benabled· Context Arena data30.8%
127Claude Opus 4.7high· Context Arena data30.8%
128MiMo V2 Pro· Context Arena data30.3%
129Qwen3.5 35B A3B· Context Arena data29.9%
130Gemini 3.1 Flash Litemedium· Context Arena data29.4%
131Qwen3.5 122B-A10B· Context Arena data28.9%
132MiMo V2.5· Context Arena data28.1%
133Step 3.7 Flashlow· Context Arena data27.8%
134Step 3.7 Flashmedium· Context Arena data27.7%
135Doubao Seed 2.1 Turbo· Context Arena data27.6%
136Gemini 3.5 Flashlow· Context Arena data27%
137GPT 5.4· Context Arena data26.9%
138GLM 5.1· Context Arena data26.9%
139DeepSeek V4 Flash· Context Arena data26.7%
140GPT-5.4 minilow· Context Arena data26%
141Gemini 3.5 Flash· Context Arena data25.9%
142DeepSeek V4 Flash 0731· Context Arena data25.6%
143MiMo V2.5enabled· Context Arena data24.8%
144Step 3.7 Flashhigh· Context Arena data24.3%
145Claude Opus 4.7xhigh· Context Arena data23.9%
146Claude Opus 4.7medium· Context Arena data23.7%
147Gemini 3 Flash· Context Arena data23.5%
148GPT 5.4 Nanomedium· Context Arena data23.3%
149Gemini 3.5 Flash Liteminimal· Context Arena data22.5%
150GPT 5.4 Nanohigh· Context Arena data22.3%

Showing 101–150 of 177 results.