Benchmark
MAEB(v1, audio-only) - Any2AnyRetrieval
Scores are percentages.
- Models measured
- 11
- Scores recorded
- 11
- Sources
- 1
- IQ weight
- 0.71
Score distribution
- 68.3%
- Mean
- 69.7%
- Median
- 9.7%
- Std dev
- 55.4%
- 10th percentile
- 82.5%
- 90th percentile
- 50.5%
- Min
This benchmark has a weight of 0.71 for speech models, measured over 5 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Jina Embeddings v5 Omni SmallAny2AnyRetrieval· MTEB Multimodal Boards data docs | 83.9% |
| 2 | Jina Embeddings v5 Omni NanoAny2AnyRetrieval· MTEB Multimodal Boards data docs | 82.5% |
| 3 | Whisper Large V3Any2AnyRetrieval· MTEB Multimodal Boards data docs | 72.9% |
| 4 | Whisper MediumAny2AnyRetrieval· MTEB Multimodal Boards data docs | 72.1% |
| 5 | Whisper SmallAny2AnyRetrieval· MTEB Multimodal Boards data docs | 70% |
| 6 | Whisper BaseAny2AnyRetrieval· MTEB Multimodal Boards data docs | 69.7% |
| 7 | Whisper TinyAny2AnyRetrieval· MTEB Multimodal Boards data docs | 69.2% |
| 8 | qwen3-omni-30b-a3b-instructAny2AnyRetrieval· MTEB Multimodal Boards data docs | 63.2% |
| 9 | Qwen2.5-Omni 7BAny2AnyRetrieval· MTEB Multimodal Boards data docs | 62.1% |
| 10 | embeddinggemma-2Any2AnyRetrieval· MTEB Multimodal Boards data docs | 55.4% |
| 11 | Qwen3 Omni 30B A3B ThinkingAny2AnyRetrieval· MTEB Multimodal Boards data docs | 50.5% |