Harrier OSS v1 27B vs Qwen 3 Embedding 8B
Harrier OSS v1 27B74.3
Qwen 3 Embedding 8B70.6
| Benchmark | Harrier OSS v1 27B | Qwen 3 Embedding 8B |
|---|---|---|
| MTEB multilingual v2 · Mean over tasks · multilingual (1037 languages) | 74.3 | 70.6 |
Specs
| Spec | Harrier OSS v1 27B | Qwen 3 Embedding 8B |
|---|---|---|
| Price | — | !$0.01 |
| Context | 33K | 33K |
| Weights | Open | Open |
| Released | Mar 2026 | Jun 2025 |
| Published speed | — | 3.3 × vLLM H100 (Baseten) |
| Dimensions | 5,376 | 4,096 |
| Max input | 33K tokens | 33K tokens |
| Languages | en, Esperanto, es +90 | — |
| Memory (4-bit) | 16 GB | 4.5 GB |
| Licence | MIT | Apache-2.0 |
| Commercial use | Allowed | Allowed |
What breaks when you switch
Harrier OSS v1 27B → Qwen 3 Embedding 8B
Re-embed your whole corpus›
Vectors from different models can't be mixed; dimensions change from 5,376 to 4,096.
Languages: not verified
Qwen 3 Embedding 8B → Harrier OSS v1 27B
Re-embed your whole corpus›
Vectors from different models can't be mixed; dimensions change from 4,096 to 5,376.