gemma-3-1b-it
by google
1.0B params · text-generation · 870 likes · 3.3M downloads
gemma-3-1b-it is a 1.0B parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (1313 tok/s), GeForce RTX 5090 (1969 tok/s), M4 Max 128GB (480 tok/s).
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 1313 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 1969 tok/s | comfortable |
| M4 Max 128GB | 128GB | 480 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 240 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 240 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 2405 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 4838 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 1160 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 960 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 931 tok/s | comfortable |