Our catalog tracks 9 DeepSeek models (7.6B–861.6 parameters). For each, the table shows the cheapest retail GPU that fully fits it at Q4 / 8K context and the fastest retail GPU, with decode tok/s. Fit means weights + KV cache + 1 GB runtime overhead within 95% of VRAM — not weights alone. 3 of them exceed every retail GPU we track and are marked accordingly. Methodology.
| # | Model | Params | Q4/8K needs (GB) | Best value GPU (fits) | Fastest retail GPU | Buy |
|---|---|---|---|---|---|---|
| 1 | DeepSeek-R1-Distill-Qwen-7B | 7.6B | 6.7 | Arc B570 10GB ≈31.8 tok/s est. | RTX 5090 32GB ≈150.0 tok/s est. | Search Amazon (affiliate) |
| 2 | DeepSeek-R1-0528-Qwen3-8B | 8.2B | 7.15 | Arc B570 10GB ≈29.5 tok/s est. | RTX 5090 32GB ≈139.1 tok/s est. | Search Amazon (affiliate) |
| 3 | DeepSeek-R1-Distill-Qwen-14B | 14.8B | 12.1 | RX 7600 XT 16GB ≈12.4 tok/s est. | RTX 4090 24GB ≈92.5 tok/s measured | Search Amazon (affiliate) |
| 4 | DeepSeek-R1-Distill-Qwen-32B | 32.8B | 25.6 | Mac mini M4 32GB ≈2.3 tok/s est. | RTX 5090 32GB ≈34.8 tok/s est. | Search Amazon (affiliate) |
| 5 | DeepSeek-R1-Distill-Llama-70B | 70.6B | 53.95 | Mac mini M4 Pro 64GB ≈2.5 tok/s est. | Mac Studio M3 Ultra 96GB ≈7.4 tok/s est. | Search Amazon (affiliate) |
| 6 | DeepSeek-V4-Flash | 158.1B MoE 13B act. | 90.55 | Mac Studio M3 Ultra 96GB ≈40.1 tok/s est. | Mac Studio M3 Ultra 96GB ≈40.1 tok/s est. | Search Amazon (affiliate) |
| 7 | DeepSeek-R1 | 684.5B MoE 37B act. | 384.88 | No retail GPU fits | — | — |
| 8 | DeepSeek-V3.2 | 685.4B MoE 37B act. | 385.37 | No retail GPU fits | — | — |
| 9 | DeepSeek-V4-Pro | 861.6B MoE 49B act. | 484.68 | No retail GPU fits | — | — |