| SPECIFICATIONS | |||||
| Max dimension | 2K | 3K | 2K | 4K | |
| Max tokens | 33K | 2K | 32K | 33K | |
| Parameters | 3.8B | — | — | 8B | |
| License | Apache 2.0 | Proprietary | Proprietary | Apache 2.0 | |
| EMBEDDING TYPES | |||||
| Types |
DenseLate
|
Dense
|
Dense
|
Dense
|
|
| MATRYOSHKA DIMENSIONS | |||||
| Available sizes |
12825651210242048
|
1282565127681024153620483072
|
25651210242048
|
3264128256512102420484096
|
|
| INPUT MODALITIES | |||||
| Text | ✓ | ✓ | ✓ | ✓ | |
| Image | ✓ | — | — | — | |
| ✓ | — | — | — | ||
| OUTPUT TYPES | |||||
| Single vector | ✓ | ✓ | ✓ | ✓ | |
| Multi vector | ✓ | — | — | — | |
| LANGUAGE SUPPORT | |||||
| Languages | 🌍 29+ | 🌍 Multi | 🌍 Multi | 🌍 100+ | |
| PRICING | |||||
| Per 1M tokens | $0.050 | $0.150 | — | — | |
| BENCHMARKS | |||||
| CMTEB CHINESE | — | — | — | 73.84 | |
| COIR | 71.59 | — | — | — | |
| JINA VDR | 84.11 | — | — | — | |
| LONGEMBED | 67.11 | — | — | — | |
| MMTEB | 66.49 | — | — | — | |
| MTEB 1536D | — | 68.17 | — | — | |
| MTEB 2048D | — | 68.16 | — | — | |
| MTEB 768D | — | 67.99 | — | — | |
| MTEB EN | 55.97 | — | — | — | |
| MTEB EN V2 | — | — | — | 75.22 | |
| MTEB MULTILINGUAL | — | — | — | 70.58 | |
| MTEB MULTILINGUAL AVG | — | 68.32 | — | — | |
| VIDORE | 90.17 | — | — | — | |