Skip to content

Benchmark

ArmBench-LLM Armenian punctuation (punctuation_accuracy, 0-shot)

Higher scores are better.

Models measured
27
Scores recorded
28
Sources
1
IQ weight
1.33

Score distribution

0.569
Mean
0.634
Median
0.202
Std dev
0.391
10th percentile
0.739
90th percentile
0
Min
00.784
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.33 for chat models, measured over 27 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1GPT-5.2 Prosetting not named by the source· ArmBench-LLM Armenian data docs0.784
2GPT-5.2 Prosetting not named by the source· ArmBench-LLM Armenian data docs0.784
3Gemini 2.5 Flash· ArmBench-LLM Armenian data docs0.747
4Gemini 3 Pro Preview· ArmBench-LLM Armenian data docs0.739
5Mistral Large 3· ArmBench-LLM Armenian data docs0.73
6Gemini 3.1 Pro Preview· ArmBench-LLM Armenian data docs0.714
7Gemma 4 31B IT· ArmBench-LLM Armenian data docs0.712
8Gemini 3 Flash· ArmBench-LLM Armenian data docs0.695
9Mistral Small 3.2 24B Instruct· ArmBench-LLM Armenian data docs0.693
10GPT-5.4 mini· ArmBench-LLM Armenian data docs0.673
11Gemini 2.5 Pro· ArmBench-LLM Armenian data docs0.657
12GPT 5.4 Nano· ArmBench-LLM Armenian data docs0.637
13Claude Sonnet 3.7· ArmBench-LLM Armenian data docs0.634
14Grok 4.1 Fast· ArmBench-LLM Armenian data docs0.634
15Gemma 3 27B IT· ArmBench-LLM Armenian data docs0.628
16Grok 4 Fast· ArmBench-LLM Armenian data docs0.618
17GLM 5· ArmBench-LLM Armenian data docs0.598
18Claude Sonnet 4· ArmBench-LLM Armenian data docs0.583
19Ministral 14B· ArmBench-LLM Armenian data docs0.576
20Gemma 3 12B IT· ArmBench-LLM Armenian data docs0.542
21Qwen3.5 27B· ArmBench-LLM Armenian data docs0.53
22Qwen3.5 35B A3B· ArmBench-LLM Armenian data docs0.528
23GLM 4.7 Flash· ArmBench-LLM Armenian data docs0.509
24Qwen3 4B Instruct 2507· ArmBench-LLM Armenian data docs0.412
25Gemma 3 4B IT· ArmBench-LLM Armenian data docs0.391
26Qwen3.5 9B· ArmBench-LLM Armenian data docs0.199
27Grok 4.20· ArmBench-LLM Armenian data docs0
28Qwen3 4B Base· ArmBench-LLM Armenian data docs0

Data credits: ArmBench-LLM by Metric-AI. Public numeric primary task results; no separate results licence asserted.