Loading…
Loading…
Models / DeepSeek-Math
GPU memory needed (smallest version of each): deepseek-math 7B about 19.8 GB, DeepSeek-Math ~685B about 784.2 GB. The cheapest runs on 2x RTX 3060 from $0.16/h right now.
2 models (7B to ~685B) · ↓ 2.7K downloads a month · from $0.16/h on 2x RTX 3060 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-math-7b-rl | vLLM | 14 GB | 19.8 GB | $0.16/h 2x RTX 3060 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originaldeepseek-ai/DeepSeek-Math-V2 | vLLM | 689 GB | 784.2 GB | $43.49/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").