Qwen 3 Embedding 8B vs Seed1.6 Embedding 1215
Qwen 3 Embedding 8B70.6
Seed1.6 Embedding 121570.3
| Benchmark | Qwen 3 Embedding 8B | Seed1.6 Embedding 1215 |
|---|---|---|
| MTEB multilingual v2 · Mean over tasks · multilingual (1037 languages) | 70.6 | 70.3 |
Specs
| Spec | Qwen 3 Embedding 8B | Seed1.6 Embedding 1215 |
|---|---|---|
| Price | !$0.01 | — |
| Context | 33K | 33K |
| Weights | Open | Closed |
| Released | Jun 2025 | Dec 2025 |
| Published speed | 3.3 × vLLM H100 (Baseten) | — |
| Dimensions | 4,096 | — |
| Max input | 33K tokens | 33K tokens |
| Memory (4-bit) | 4.5 GB | — |
| Licence | Apache-2.0 | — |
| Commercial use | Allowed | — |
What breaks when you switch
Qwen 3 Embedding 8B → Seed1.6 Embedding 1215
Re-embed your whole corpus›
Vectors from different models can't be mixed.
Seed1.6 Embedding 1215 → Qwen 3 Embedding 8B
Re-embed your whole corpus›
Vectors from different models can't be mixed.