Qwen2.5-0.5B
by Qwen
494M params · text-generation · 424 likes · 2.4M downloads
Qwen2.5-0.5B is a 494M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (2657 tok/s), GeForce RTX 5090 (3985 tok/s), M4 Max 128GB (972 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| Featherless |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 2657 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 3985 tok/s | comfortable |
| M4 Max 128GB | 128GB | 972 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 486 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 486 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 4869 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 9793 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 2349 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 1943 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 1886 tok/s | comfortable |