Models / granite-4.1
granite-4.1 by ibm-granite
6 models (3B to 30B) · ↓ 1.1M downloads a month · from $0.08/h on GTX 1660 SUPER right now. Each model's versions do the same job; quantized ones (AWQ, FP8, 4-bit) need less GPU memory and run on cheaper machines.
granite-4.1 3B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originalibm-granite/granite-4.1-3b | vLLM | 6.8 GB | 10.5 GB | $0.10/h RTX 3060 |
| Onnx (Prince-1)Prince-1/Granite-4.1-3b-Onnx | ONNX Runtime GenAI | 6.8 GB | 10 GB | $0.09/h RTX 3060 |
| Onnx (onnx-community)onnx-community/Granite-4.1-3b-Onnx | ONNX Runtime GenAI | 6.8 GB | 10 GB | $0.09/h RTX 3060 |
| Onnx q4 (MiCkSoftware)MiCkSoftware/Granite-4.1-3b-Onnx-q4 | ONNX Runtime GenAI | 2.2 GB | 4.7 GB | $0.08/h GTX 1660 SUPER |
granite-4.1 3B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originalibm-granite/granite-4.1-3b-base | vLLM | 6.8 GB | 10.5 GB | $0.10/h RTX 3060 |
| granite 4.2 3bibm-granite/granite-4.2-3b | vLLM | 7.3 GB | 11.1 GB | $0.10/h RTX 3060 |
granite-4.1 8B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originalibm-granite/granite-4.1-8b | vLLM | 18 GB | 23.3 GB | $0.24/h RTX 3090 |
| fp8ibm-granite/granite-4.1-8b-fp8 | vLLM | 9.6 GB | 14.4 GB | $0.14/h RTX A4000 |
| granite guardian 4.1 8bibm-granite/granite-guardian-4.1-8b | vLLM | 17 GB | 22.3 GB | $0.24/h RTX 3090 |
| Onnx (Prince-1)Prince-1/Granite-4.1-8b-Onnx | ONNX Runtime GenAI | 17 GB | 22.2 GB | $0.18/h Nvidia V100 |
| Onnx (onnx-community)onnx-community/Granite-4.1-8b-Onnx | ONNX Runtime GenAI | 17 GB | 22.2 GB | $0.18/h Nvidia V100 |
granite-4.1 8B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| granite 4.2 8bibm-granite/granite-4.2-8b | vLLM | 18 GB | 23.3 GB | $0.24/h RTX 3090 |
granite-4.1 30B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originalibm-granite/granite-4.1-30b | vLLM | 58 GB | 68.8 GB | $0.89/h 2x Nvidia A40 |
| Onnx (Prince-1)Prince-1/Granite-4.1-30b-Onnx | ONNX Runtime GenAI | 58 GB | 70.1 GB | $1.04/h Nvidia A100 |
granite-4.1 30B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| granite 4.2 30bibm-granite/granite-4.2-30b | vLLM | 59 GB | 69.7 GB | $0.89/h 2x Nvidia A40 |
Model details come from Hugging Face and are refreshed daily; prices follow the live offers. GPU memory is GridPup's estimate: the weights, room for an 8,192-token conversation (chat models) and the runtime's own memory. Large models are split across the GPUs of one machine ("4x RTX 4090").