Skip to content

Benchmark

KazBench Kazakh translation (chrF, dev)

Higher scores are better.

Models measured
6
Scores recorded
6
Sources
1
IQ weight
1.12

Score distribution

77.5
Mean
88.2
Median
23.2
Std dev
79.8
10th percentile
92.3
90th percentile
26.4
Min
26.492.3
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.12 for chat models, measured over 6 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-4o minidev, version: 0.1.0· KazBench Kazakh data docs92.3
2Llama 4 Scoutdev, version: 0.1.0· KazBench Kazakh data docs92.1
3Gemini 2.5 Flashdev, version: 0.1.0· KazBench Kazakh data docs88.2
4Claude Haiku 3.5dev, version: 0.1.0· KazBench Kazakh data docs86.1
5Qwen2.5 72B Instructdev, version: 0.1.0· KazBench Kazakh data docs79.8
6Llama 3.1 8B Instantdev, version: 0.1.0· KazBench Kazakh data docs26.4

Data credits: KazBench by Yersultan Akhmer. Provisional DEV numeric results; code MIT and task-data CC BY 4.0 do not establish a separate score licence.