Top-ranked embedding model on MTEB English and Chinese. Built on Qwen2-7B with bidirectional attention and instruction tuning.
Modalities
Dimensions
4K
Max tokens
32K
Parameters
7B
Price / 1M tokens
—
Type
🌍 Multilingual support
| MTEB 56 | 70.24 |
| CMTEB 35 | 72.05 |
| MTEB FR | 68.25 |
| MTEB PL | 67.86 |
| Release date | 2024-06-16 |
| License | Apache 2.0 |
| Model ID | gte-qwen2-7b-instruct |
| Provider | Alibaba |