snowflake-arctic-embed-m
by Snowflake
109M params · sentence-similarity · 163 likes · 384.0k downloads
snowflake-arctic-embed-m is a 109M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (12057 tok/s), GeForce RTX 5090 (18082 tok/s), M4 Max 128GB (4412 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| HF Inference |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 12057 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 18082 tok/s | comfortable |
| M4 Max 128GB | 128GB | 4412 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 2206 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 2206 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 22091 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 44433 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 10660 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 8816 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 8557 tok/s | comfortable |