gte-large-zh
by thenlper
326M params · sentence-similarity · 121 likes · 5.4k downloads
gte-large-zh is a 326M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (4033 tok/s), GeForce RTX 5090 (6048 tok/s), M4 Max 128GB (1476 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| HF Inference |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 4033 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 6048 tok/s | comfortable |
| M4 Max 128GB | 128GB | 1476 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 738 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 738 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 7389 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 14863 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 3566 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 2949 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 2862 tok/s | comfortable |