Loading…
Loading…
Models / ERNIE-4.5
3 models (0.3B to 21B) · ↓ 114.3K 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.
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originalbaidu/ERNIE-4.5-0.3B-PT | vLLM | 720 MB | 3.2 GB | $0.10/h RTX 3060 |
| ERNIE 4.5 0.3B Onnx (Prince-1)Prince-1/ERNIE-4.5-0.3B-Onnx | ONNX Runtime GenAI | 770 MB | 2.5 GB | $0.08/h GTX 1660 SUPER |
| ERNIE 4.5 0.3B Onnx (fireicesion)fireicesion/ERNIE-4.5-0.3B-Onnx | ONNX Runtime GenAI | 770 MB | 2.5 GB | $0.08/h GTX 1660 SUPER |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originalbaidu/ERNIE-4.5-21B-A3B-Thinking | vLLM | 44 GB | 51.3 GB | $0.75/h 4x RTX 5060 Ti |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originalbaidu/ERNIE-4.5-21B-A3B-PT | vLLM | 44 GB | 51.5 GB | $0.75/h 4x RTX 5060 Ti |
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").