Qwen3 Embedding 8B vs Cohere Embed Multilingual v3
Detailed comparison between Qwen3 Embedding 8B and Cohere Embed Multilingual v3. See which embedding best meets your accuracy and performance needs.
Model Comparison
Qwen3 Embedding 8B takes the lead.
Both Qwen3 Embedding 8B and Cohere Embed Multilingual v3 are powerful embedding models designed to improve retrieval quality in RAG applications. However, their performance characteristics differ in important ways.
Why Qwen3 Embedding 8B:
- Qwen3 Embedding 8B has 15 higher ELO rating
- Qwen3 Embedding 8B delivers better accuracy (nDCG@10: 0.818 vs 0.781)
- Cohere Embed Multilingual v3 is 106734ms faster on average
- Qwen3 Embedding 8B has a 5.1% higher win rate
Overview
Key metrics
ELO Rating
Overall ranking quality
Qwen3 Embedding 8B
Cohere Embed Multilingual v3
Win Rate
Head-to-head performance
Qwen3 Embedding 8B
Cohere Embed Multilingual v3
Accuracy (nDCG@10)
Ranking quality metric
Qwen3 Embedding 8B
Cohere Embed Multilingual v3
Average Latency
Response time
Qwen3 Embedding 8B
Cohere Embed Multilingual v3
Visual Performance Analysis
Performance
ELO Rating Comparison
Win/Loss/Tie Breakdown
Accuracy Across Datasets (nDCG@10)
Latency Distribution (ms)
Breakdown
How the models stack up
| Metric | Qwen3 Embedding 8B | Cohere Embed Multilingual v3 | Description |
|---|---|---|---|
| Overall Performance | |||
| ELO Rating | 1516 | 1501 | Overall ranking quality based on pairwise comparisons |
| Win Rate | 47.9% | 42.9% | Percentage of comparisons won against other models |
| Pricing & Availability | |||
| Price per 1M tokens | $0.050 | $0.100 | Cost per million tokens processed |
| Release Date | 2025-06-06 | 2024-02-07 | Model release date |
| Accuracy Metrics | |||
| Avg nDCG@10 | 0.818 | 0.781 | Normalized discounted cumulative gain at position 10 |
| Performance Metrics | |||
| Avg Latency | 130758ms | 24024ms | Average response time across all datasets |
Dataset Performance
By field
Comprehensive comparison of accuracy metrics (nDCG, Recall) and latency percentiles for each benchmark dataset.
PG
| Metric | Qwen3 Embedding 8B | Cohere Embed Multilingual v3 | Description |
|---|---|---|---|
| Accuracy Metrics | |||
| Latency Metrics | |||
| Mean | 151470ms | 37785ms | Average response time |
| P50 | 148441ms | 37029ms | 50th percentile (median) |
| P90 | 174191ms | 43453ms | 90th percentile |
business reports
| Metric | Qwen3 Embedding 8B | Cohere Embed Multilingual v3 | Description |
|---|---|---|---|
| Accuracy Metrics | |||
| Latency Metrics | |||
| Mean | 93874ms | 5204ms | Average response time |
| P50 | 91997ms | 5100ms | 50th percentile (median) |
| P90 | 107955ms | 5985ms | 90th percentile |
DBPedia
| Metric | Qwen3 Embedding 8B | Cohere Embed Multilingual v3 | Description |
|---|---|---|---|
| Accuracy Metrics | |||
| nDCG@5 | 0.633 | 0.619 | Ranking quality at top 5 results |
| nDCG@10 | 0.625 | 0.591 | Ranking quality at top 10 results |
| Recall@5 | 0.235 | 0.222 | % of relevant docs in top 5 |
| Recall@10 | 0.381 | 0.329 | % of relevant docs in top 10 |
| Latency Metrics | |||
| Mean | 200373ms | 33570ms | Average response time |
| P50 | 196366ms | 32899ms | 50th percentile (median) |
| P90 | 230429ms | 38606ms | 90th percentile |
FiQa
| Metric | Qwen3 Embedding 8B | Cohere Embed Multilingual v3 | Description |
|---|---|---|---|
| Accuracy Metrics | |||
| nDCG@5 | 0.760 | 0.637 | Ranking quality at top 5 results |
| nDCG@10 | 0.781 | 0.654 | Ranking quality at top 10 results |
| Recall@5 | 0.732 | 0.621 | % of relevant docs in top 5 |
| Recall@10 | 0.814 | 0.692 | % of relevant docs in top 10 |
| Latency Metrics | |||
| Mean | 189471ms | 37581ms | Average response time |
| P50 | 185682ms | 36829ms | 50th percentile (median) |
| P90 | 217892ms | 43218ms | 90th percentile |
SciFact
| Metric | Qwen3 Embedding 8B | Cohere Embed Multilingual v3 | Description |
|---|---|---|---|
| Accuracy Metrics | |||
| nDCG@5 | 0.750 | 0.723 | Ranking quality at top 5 results |
| nDCG@10 | 0.765 | 0.732 | Ranking quality at top 10 results |
| Recall@5 | 0.843 | 0.808 | % of relevant docs in top 5 |
| Recall@10 | 0.883 | 0.833 | % of relevant docs in top 10 |
| Latency Metrics | |||
| Mean | 200031ms | 39542ms | Average response time |
| P50 | 196030ms | 38751ms | 50th percentile (median) |
| P90 | 230036ms | 45473ms | 90th percentile |
MSMARCO
| Metric | Qwen3 Embedding 8B | Cohere Embed Multilingual v3 | Description |
|---|---|---|---|
| Accuracy Metrics | |||
| nDCG@5 | 1.000 | 0.996 | Ranking quality at top 5 results |
| nDCG@10 | 1.000 | 0.994 | Ranking quality at top 10 results |
| Recall@5 | 0.123 | 0.122 | % of relevant docs in top 5 |
| Recall@10 | 0.224 | 0.218 | % of relevant docs in top 10 |
| Latency Metrics | |||
| Mean | 180144ms | 32380ms | Average response time |
| P50 | 176541ms | 31732ms | 50th percentile (median) |
| P90 | 207166ms | 37237ms | 90th percentile |
NorQuAD
| Metric | Qwen3 Embedding 8B | Cohere Embed Multilingual v3 | Description |
|---|---|---|---|
| Accuracy Metrics | |||
| Latency Metrics | |||
| Mean | 18544ms | 3824ms | Average response time |
| P50 | 18173ms | 3748ms | 50th percentile (median) |
| P90 | 21326ms | 4398ms | 90th percentile |
ARCD
| Metric | Qwen3 Embedding 8B | Cohere Embed Multilingual v3 | Description |
|---|---|---|---|
| Accuracy Metrics | |||
| nDCG@5 | 0.913 | 0.925 | Ranking quality at top 5 results |
| nDCG@10 | 0.919 | 0.933 | Ranking quality at top 10 results |
| Recall@5 | 0.920 | 0.940 | % of relevant docs in top 5 |
| Recall@10 | 0.940 | 0.960 | % of relevant docs in top 10 |
| Latency Metrics | |||
| Mean | 12156ms | 2305ms | Average response time |
| P50 | 11913ms | 2259ms | 50th percentile (median) |
| P90 | 13979ms | 2651ms | 90th percentile |
Explore More
Compare more embeddings
See how all embedding models stack up. Compare OpenAI, Cohere, Jina AI, Voyage, and more. View comprehensive benchmarks, compare performance metrics, and find the perfect embedding for your RAG application.