Loading…
Loading…
Models / DeepSeek
5 models (~236B to ~685B) · ↓ 1.8M downloads a month · from $11.77/h on 8x RTX PRO 6000 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 |
|---|---|---|---|---|
| Originaldeepseek-ai/DeepSeek-V2-Chat | vLLM | 471 GB | 537.3 GB | $11.77/h 8x RTX PRO 6000 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originaldeepseek-ai/DeepSeek-V2 | vLLM | 471 GB | 537.3 GB | $11.77/h 8x RTX PRO 6000 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originaldeepseek-ai/DeepSeek-V2.5 | vLLM | 471 GB | 537.3 GB | $11.77/h 8x RTX PRO 6000 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originaldeepseek-ai/DeepSeek-V3 | vLLM | 689 GB | 783.2 GB | $44.03/h 8x Nvidia H200 |
| Main (unsloth)unsloth/DeepSeek-V3 | vLLM | 689 GB | 783.2 GB | $44.03/h 8x Nvidia H200 |
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").