Benchmark
Agentset Embedding Recall@10 - ARCD
Embedding benchmarks measure how well models find related items, for example when matching a search to relevant results. Scores are percentages. Higher is better.
- Models measured
- 15
- Scores recorded
- 15
- Sources
- 1
- IQ weight
- 0.50
Score distribution
- 89.5%
- Mean
- 96%
- Median
- 22.4%
- Std dev
- 90%
- 10th percentile
- 98%
- 90th percentile
- 6%
- Min
This benchmark has a weight of 0.50 for embedding models, measured over 15 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 | voyage-3-largeARCD· Agentset | 98% |
| 2 | voyage-3.5ARCD· Agentset | 98% |
| 3 | voyage-3.5-liteARCD· Agentset | 98% |
| 4 | voyage-4ARCD· Agentset | 98% |
| 5 | BGE M3ARCD· Agentset | 96% |
| 6 | Embed Multilingual V3.0ARCD· Agentset | 96% |
| 7 | Gemini Embedding 2ARCD· Agentset | 96% |
| 8 | Jina Embeddings v3ARCD· Agentset | 96% |
| 9 | Qwen3 Embedding 4BARCD· Agentset | 96% |
| 10 | Jina Embeddings v5 Text SmallARCD· Agentset | 94% |
| 11 | Qwen3 Embedding 8BARCD· Agentset | 94% |
| 12 | Text Embedding 3 LargeARCD· Agentset | 94% |
| 13 | Text Embedding 3 SmallARCD· Agentset | 92% |
| 14 | Qwen3 Embedding 0.6BARCD· Agentset | 90% |
| 15 | Text Embedding 004ARCD· Agentset | 6% |