harrier-oss-v1-270m
by microsoft
268M params · feature-extraction · 174 likes · 166.0k downloads
harrier-oss-v1-270m is a 268M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (4897 tok/s), GeForce RTX 5090 (7344 tok/s), M4 Max 128GB (1792 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| HF Inference |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 4897 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 7344 tok/s | comfortable |
| M4 Max 128GB | 128GB | 1792 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 896 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 896 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 8972 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 18047 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 4330 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 3580 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 3475 tok/s | comfortable |