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
101Qwen3.5 35B A3Benabled· Context Arena data31.9%
102MiMo V2.5 Proenabled· Context Arena data31.7%
103Kimi K2.5· Context Arena data31.6%
104Kimi K3low· Context Arena data31.1%
105GPT 5.4 Nanoxhigh· Context Arena data30.4%
106Hy3 (Deep Infra)high· Context Arena data30.1%
107GPT-5.4 minihigh· Context Arena data30%
108GPT 5.4· Context Arena data29%
109Qwen3.5 Flash· Context Arena data28.6%
110Gemini 3 Flashmedium· Context Arena data28.4%
111GPT 5.5· Context Arena data28.3%
112Gemini 3.5 Flash Liteminimal· Context Arena data28.2%
113Gemini 3.1 Flash Litehigh· Context Arena data28.1%
114Gemini 3 Flashlow· Context Arena data28.1%
115GLM 5.1enabled· Context Arena data27.4%
116MiMo V2.5enabled· Context Arena data27.2%
117MiMo V2 Proenabled· Context Arena data26.4%
118Gemini 3.1 Flash Litelow· Context Arena data26%
119Qwen3.5 27B· Context Arena data25.4%
120DeepSeek V4 Pro· Context Arena data25.3%
121Hy3 (Deep Infra)low· Context Arena data25.2%
122Nemotron 3 Super 120B A12B· Context Arena data24.9%
123Qwen3.5 122B-A10B· Context Arena data24.5%
124MiMo V2.5· Context Arena data23.7%
125Doubao Seed 2.1 Turboenabled· Context Arena data23.7%
126MiMo V2.5 Pro· Context Arena data23.5%
127Gemini 3.6 Flashminimal· Context Arena data23.4%
128Gemini 3.1 Flash Litemedium· Context Arena data23.4%
129Gemini 3.5 Flash· Context Arena data23.2%
130GPT-5.4 minilow· Context Arena data23.2%
131DeepSeek V4 Pro 0813· Context Arena data22.6%
132Qwen3.5 Plus· Context Arena data22.3%
133GPT 5.4 Nanohigh· Context Arena data22.2%
134Qwen3.5 397B A17B· Context Arena data22.1%
135Step 3.7 Flashmedium· Context Arena data21.8%
136GPT 5.4 Nanomedium· Context Arena data21.6%
137Gemini 3.5 Flashlow· Context Arena data21.6%
138Gemini 3 Flash· Context Arena data20.8%
139Qwen3.5 35B A3B· Context Arena data20.3%
140MiMo V2 Pro· Context Arena data19.6%
141GLM 5.1· Context Arena data19.6%
142Grok 4.20enabled· Context Arena data19.4%
143GPT-5.4 mini· Context Arena data18.9%
144Gemini 3.1 Flash Lite· Context Arena data18.8%
145DeepSeek V4 Flash 0731· Context Arena data17.9%
146DeepSeek V4 Flash· Context Arena data17.7%
147MiniMax M3enabled· Context Arena data17.4%
148MiniMax M2.7enabled· Context Arena data16.8%
149Step 3.7 Flashlow· Context Arena data16.8%
150Doubao Seed 2.1 Turbo· Context Arena data16.4%

Showing 101–150 of 177 results.