Skip to content

Benchmark

MLEB - Australian Tax Guidance Retrieval

Embedding benchmarks measure how well models find related items, for example when matching a search to relevant results. Scores are percentages.

Models measured
18
Scores recorded
18
Sources
1
IQ weight
0.69

Score distribution

75.3%
Mean
76.2%
Median
3.3%
Std dev
71%
10th percentile
79.9%
90th percentile
70.3%
Min
70.3%80.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.69 for embedding models, measured over 18 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Qwen3 Embedding 8BAustralian Tax Guidance Retrieval· MLEB data80.4%
2voyage-4-largeAustralian Tax Guidance Retrieval· MLEB data80%
3Gemini Embedding 001Australian Tax Guidance Retrieval· MLEB data79.9%
4Text Embedding 3 LargeAustralian Tax Guidance Retrieval· MLEB data77.8%
5Jina Embeddings v4Australian Tax Guidance Retrieval· MLEB data77.6%
6Qwen3 Embedding 4BAustralian Tax Guidance Retrieval· MLEB data77.3%
7voyage-4Australian Tax Guidance Retrieval· MLEB data77.2%
8Jina Embeddings v5 Text SmallAustralian Tax Guidance Retrieval· MLEB data76.8%
9voyage-4-liteAustralian Tax Guidance Retrieval· MLEB data76.2%
10snowflake-arctic-embed-m-v2.0Australian Tax Guidance Retrieval· MLEB data75.4%
11Text Embedding 3 SmallAustralian Tax Guidance Retrieval· MLEB data74.3%
12Jina Embeddings v5 Text NanoAustralian Tax Guidance Retrieval· MLEB data74.1%
13Qwen3 Embedding 0.6BAustralian Tax Guidance Retrieval· MLEB data72.5%
14snowflake-arctic-embed-l-v2.0Australian Tax Guidance Retrieval· MLEB data72.4%
15Text Embedding Ada 002Australian Tax Guidance Retrieval· MLEB data71.3%
16Granite Embedding Small English R2Australian Tax Guidance Retrieval· MLEB data71%
17BGE M3Australian Tax Guidance Retrieval· MLEB data70.5%
18Multilingual E5 Large InstructAustralian Tax Guidance Retrieval· MLEB data70.3%

Data credits: MLEB (Isaacus), MIT (licence)