Loading…
Loading…
Models / DeepSeek-R1-Distill-Qwen
GPU memory needed (smallest version of each): DeepSeek-R1-Distill-Qwen 14B about 36.8 GB, DeepSeek-R1-Distill-Qwen 32B about 77.4 GB. The cheapest runs on 2x RTX 3090 from $0.46/h right now.
2 models (14B to 32B) · ↓ 2K downloads a month · from $0.46/h on 2x RTX 3090 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 |
|---|---|---|---|---|
| Originalhuihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2 | vLLM | 30 GB | 36.8 GB | $0.46/h 2x RTX 3090 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originalhuihui-ai/DeepSeek-R1-Distill-Qwen-32B-abliterated | vLLM | 66 GB | 77.4 GB | $0.89/h Nvidia A800 |
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").