pygmalion-6b
by PygmalionAI
6.0B params · text-generation · 752 likes · 1.9k downloads
pygmalion-6b is a 6.0B parameter model. At Q4 quantization it requires 3GB of VRAM. It runs comfortably on GeForce RTX 4090 (218 tok/s), GeForce RTX 5090 (328 tok/s), M4 Max 128GB (80 tok/s).
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 218 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 328 tok/s | comfortable |
| M4 Max 128GB | 128GB | 80 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 40 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 40 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 400 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 806 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 193 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 160 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 155 tok/s | comfortable |