VRAM usage for all quantization and cache format combinations. Base overhead: 1.5 GB (CUDA context + activations).
| Quantization | Cache Format | Model Weights | 8K Context | 16K Context | 32K Context | 40K Context |
|---|---|---|---|---|---|---|
| FP16 16.0 bpw | FP32 | 493.5 GB | 497.94 GB (+2.94 KV) | 500.88 GB (+5.88 KV) | 506.75 GB (+11.75 KV) | 509.69 GB (+14.69 KV) |
| FP16 16.0 bpw | FP16 | 493.5 GB | 496.47 GB (+1.47 KV) | 497.94 GB (+2.94 KV) | 500.88 GB (+5.88 KV) | 502.34 GB (+7.34 KV) |
| FP16 16.0 bpw | Q8_0 | 493.5 GB | 495.81 GB (+0.81 KV) | 496.62 GB (+1.62 KV) | 498.23 GB (+3.23 KV) | 499.04 GB (+4.04 KV) |
| FP16 16.0 bpw | FP8 (Exp) | 493.5 GB | 495.73 GB (+0.73 KV) | 496.47 GB (+1.47 KV) | 497.94 GB (+2.94 KV) | 498.67 GB (+3.67 KV) |
| FP16 16.0 bpw | Q4_0 (Exp) | 493.5 GB | 495.44 GB (+0.44 KV) | 495.88 GB (+0.88 KV) | 496.76 GB (+1.76 KV) | 497.2 GB (+2.2 KV) |
| Q8_0 8.0 bpw | FP32 | 246.75 GB | 251.19 GB (+2.94 KV) | 254.12 GB (+5.88 KV) | 260.0 GB (+11.75 KV) | 262.94 GB (+14.69 KV) |
| Q8_0 8.0 bpw | FP16 | 246.75 GB | 249.72 GB (+1.47 KV) | 251.19 GB (+2.94 KV) | 254.12 GB (+5.88 KV) | 255.59 GB (+7.34 KV) |
| Q8_0 8.0 bpw | Q8_0 | 246.75 GB | 249.06 GB (+0.81 KV) | 249.87 GB (+1.62 KV) | 251.48 GB (+3.23 KV) | 252.29 GB (+4.04 KV) |
| Q8_0 8.0 bpw | FP8 (Exp) | 246.75 GB | 248.98 GB (+0.73 KV) | 249.72 GB (+1.47 KV) | 251.19 GB (+2.94 KV) | 251.92 GB (+3.67 KV) |
| Q8_0 8.0 bpw | Q4_0 (Exp) | 246.75 GB | 248.69 GB (+0.44 KV) | 249.13 GB (+0.88 KV) | 250.01 GB (+1.76 KV) | 250.45 GB (+2.2 KV) |
| Q4_K_M 4.65 bpw | FP32 | 143.42 GB | 147.86 GB (+2.94 KV) | 150.8 GB (+5.88 KV) | 156.67 GB (+11.75 KV) | 159.61 GB (+14.69 KV) |
| Q4_K_M 4.65 bpw | FP16 | 143.42 GB | 146.39 GB (+1.47 KV) | 147.86 GB (+2.94 KV) | 150.8 GB (+5.88 KV) | 152.27 GB (+7.34 KV) |
| Q4_K_M 4.65 bpw | Q8_0 | 143.42 GB | 145.73 GB (+0.81 KV) | 146.54 GB (+1.62 KV) | 148.15 GB (+3.23 KV) | 148.96 GB (+4.04 KV) |
| Q4_K_M 4.65 bpw | FP8 (Exp) | 143.42 GB | 145.66 GB (+0.73 KV) | 146.39 GB (+1.47 KV) | 147.86 GB (+2.94 KV) | 148.6 GB (+3.67 KV) |
| Q4_K_M 4.65 bpw | Q4_0 (Exp) | 143.42 GB | 145.36 GB (+0.44 KV) | 145.8 GB (+0.88 KV) | 146.69 GB (+1.76 KV) | 147.13 GB (+2.2 KV) |
| Q4_K_S 4.58 bpw | FP32 | 141.26 GB | 145.7 GB (+2.94 KV) | 148.64 GB (+5.88 KV) | 154.51 GB (+11.75 KV) | 157.45 GB (+14.69 KV) |
| Q4_K_S 4.58 bpw | FP16 | 141.26 GB | 144.23 GB (+1.47 KV) | 145.7 GB (+2.94 KV) | 148.64 GB (+5.88 KV) | 150.11 GB (+7.34 KV) |
| Q4_K_S 4.58 bpw | Q8_0 | 141.26 GB | 143.57 GB (+0.81 KV) | 144.38 GB (+1.62 KV) | 146.0 GB (+3.23 KV) | 146.8 GB (+4.04 KV) |
| Q4_K_S 4.58 bpw | FP8 (Exp) | 141.26 GB | 143.5 GB (+0.73 KV) | 144.23 GB (+1.47 KV) | 145.7 GB (+2.94 KV) | 146.44 GB (+3.67 KV) |
| Q4_K_S 4.58 bpw | Q4_0 (Exp) | 141.26 GB | 143.21 GB (+0.44 KV) | 143.65 GB (+0.88 KV) | 144.53 GB (+1.76 KV) | 144.97 GB (+2.2 KV) |
| Q3_K_M 3.91 bpw | FP32 | 120.6 GB | 125.04 GB (+2.94 KV) | 127.97 GB (+5.88 KV) | 133.85 GB (+11.75 KV) | 136.79 GB (+14.69 KV) |
| Q3_K_M 3.91 bpw | FP16 | 120.6 GB | 123.57 GB (+1.47 KV) | 125.04 GB (+2.94 KV) | 127.97 GB (+5.88 KV) | 129.44 GB (+7.34 KV) |
| Q3_K_M 3.91 bpw | Q8_0 | 120.6 GB | 122.91 GB (+0.81 KV) | 123.71 GB (+1.62 KV) | 125.33 GB (+3.23 KV) | 126.14 GB (+4.04 KV) |
| Q3_K_M 3.91 bpw | FP8 (Exp) | 120.6 GB | 122.83 GB (+0.73 KV) | 123.57 GB (+1.47 KV) | 125.04 GB (+2.94 KV) | 125.77 GB (+3.67 KV) |
| Q3_K_M 3.91 bpw | Q4_0 (Exp) | 120.6 GB | 122.54 GB (+0.44 KV) | 122.98 GB (+0.88 KV) | 123.86 GB (+1.76 KV) | 124.3 GB (+2.2 KV) |
| Q2_K 2.63 bpw | FP32 | 81.12 GB | 85.56 GB (+2.94 KV) | 88.49 GB (+5.88 KV) | 94.37 GB (+11.75 KV) | 97.31 GB (+14.69 KV) |
| Q2_K 2.63 bpw | FP16 | 81.12 GB | 84.09 GB (+1.47 KV) | 85.56 GB (+2.94 KV) | 88.49 GB (+5.88 KV) | 89.96 GB (+7.34 KV) |
| Q2_K 2.63 bpw | Q8_0 | 81.12 GB | 83.43 GB (+0.81 KV) | 84.23 GB (+1.62 KV) | 85.85 GB (+3.23 KV) | 86.66 GB (+4.04 KV) |
| Q2_K 2.63 bpw | FP8 (Exp) | 81.12 GB | 83.35 GB (+0.73 KV) | 84.09 GB (+1.47 KV) | 85.56 GB (+2.94 KV) | 86.29 GB (+3.67 KV) |
| Q2_K 2.63 bpw | Q4_0 (Exp) | 81.12 GB | 83.06 GB (+0.44 KV) | 83.5 GB (+0.88 KV) | 84.38 GB (+1.76 KV) | 84.82 GB (+2.2 KV) |
Total VRAM = Model Weights + KV Cache + 1.5 GB overhead. Actual usage may vary ±5% based on inference engine and optimizations.
Use our calculator to see if this model fits your specific hardware configuration.