Qwen2.5-14B-Instruct-1M
by Qwen
14.8B params · text-generation · 334 likes · 50.3k downloads
Qwen2.5-14B-Instruct-1M is a 14.8B parameter model. At Q4 quantization it requires 7GB of VRAM. It runs comfortably on GeForce RTX 4090 (88 tok/s), GeForce RTX 5090 (133 tok/s), M4 Max 128GB (32 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| Featherless |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 88 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 133 tok/s | comfortable |
| M4 Max 128GB | 128GB | 32 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 16 tok/s | tight |
| M4 Pro 24GB | 24GB | 16 tok/s | tight |
| A100 PCIe 80 GB | 80GB | 162 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 327 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 78 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 64 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 63 tok/s | comfortable |