bge-small-en-v1.5
by BAAI
33M params · feature-extraction · 419 likes · 6.8M downloads
bge-small-en-v1.5 is a 33M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (39357 tok/s), GeForce RTX 5090 (59021 tok/s), M4 Max 128GB (14402 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| HF Inference |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 39357 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 59021 tok/s | comfortable |
| M4 Max 128GB | 128GB | 14402 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 7201 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 7201 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 72109 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 145033 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 34798 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 28776 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 27932 tok/s | comfortable |