Loading…
Loading…
Models / DeepSeek-V4-Flash
1 model (~291B) · ↓ 828.5K downloads a month · from $2.36/h on 4x Nvidia A40 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-V4-Flash | vLLM | 160 GB | 180.4 GB | $2.36/h 4x Nvidia A40 |
| NVFP4 (nvidia)nvidia/DeepSeek-V4-Flash-NVFP4 | vLLM | 168 GB | 190 GB | $4.73/h 8x RTX 5090 |
| MXFP4 (amd)amd/DeepSeek-V4-Flash-MXFP4 | vLLM | 159 GB | 179.8 GB | $2.36/h 4x 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").