Skip to content

Benchmark

LaoBench Lao-7k translation Environment & Development BLEU (zero-shot)

Multilingual benchmarks compare model performance across languages. Higher scores are better.

Models and systems measured
12
Scores recorded
13
Sources
1
IQ weight
1.25

Score distribution

26.6
Mean
28.4
Median
7.5
Std dev
23
10th percentile
32.4
90th percentile
2.1
Min
2.133.7
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.25 for chat models, measured over 10 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model or systemScore
1Gemini 2.5 Protable: 3, mode: single· LaoBench (arXiv:2511.11334v3) data33.7
2Claude Opus 4.1table: 3, mode: single· LaoBench (arXiv:2511.11334v3) data32.4
3DeepSeek V3.2 Exptable: 3, mode: Non-Thinking· LaoBench (arXiv:2511.11334v3) data29.5
4Qwen3 235B A22B Instruct 2507table: 3, mode: single· LaoBench (arXiv:2511.11334v3) data29.1
5Qwen3 Maxtable: 3, mode: single· LaoBench (arXiv:2511.11334v3) data28.9
6Claude Sonnet 4.5table: 3, mode: single· LaoBench (arXiv:2511.11334v3) data28.4
7Qwen 3 235B A22Btable: 3, mode: single· LaoBench (arXiv:2511.11334v3) data28
8DeepSeek V3.2 Exptable: 3, mode: Thinking· LaoBench (arXiv:2511.11334v3) data27.3
9GPT 5table: 3, mode: single· LaoBench (arXiv:2511.11334v3) data26.9
10Qwen3 Next 80B A3B Instructtable: 3, mode: single· LaoBench (arXiv:2511.11334v3) data23
11Ling Mini 2.0table: 3, mode: single· LaoBench (arXiv:2511.11334v3) data2.1

Not ranked

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
Qwen3-PlusNon-Thinkingtable: 3, mode: Non-Thinking· LaoBench (arXiv:2511.11334v3) data28.9
Qwen3-PlusThinkingtable: 3, mode: Thinking· LaoBench (arXiv:2511.11334v3) data28.3

Data credits: LaoBench, BAAI et al., arXiv:2511.11334v3. Cited numeric results only. arXiv non-exclusive distribution notice establishes no results reuse grant; dataset Apache licence is not a results grant.