Can I run MiMo-V2-Flash on a Apple M5 Max?
Not at these settings. No indexed quantization of MiMo-V2-Flash fits Apple M5 Max at any context we compute, with q4_0 KV. The smallest shipped quantization is 58.20 GiB in weights alone, against 25.11 GiB usable. CPU offload can still run it, slowly.
Every quantization at every context
| Quant | Weights● | 4K◐ | 8K◐ | 16K◐ | 32K◐ | 64K◐ | 128K◐ |
|---|---|---|---|---|---|---|---|
| BF16 | 575.24 GiB | 576.0 | 576.1 | 576.4 | 576.9 | 578.0 | 580.1 |
| Q8_0 | 305.69 GiB | 306.4 | 306.5 | 306.8 | 307.3 | 308.4 | 310.5 |
| Q6_K | 236.11 GiB | 236.8 | 237.0 | 237.2 | 237.8 | 238.8 | 240.9 |
| Q5_K_M | 204.17 GiB | 204.9 | 205.0 | 205.3 | 205.8 | 206.9 | 209.0 |
| Q5_K_S | 198.04 GiB | 198.8 | 198.9 | 199.2 | 199.7 | 200.7 | 202.9 |
| Q4_1 | 180.12 GiB | 180.9 | 181.0 | 181.2 | 181.8 | 182.8 | 184.9 |
| I1-Q4_1 | 180.01 GiB | 180.7 | 180.9 | 181.1 | 181.7 | 182.7 | 184.8 |
| Q4_K_L | 174.72 GiB | 175.4 | 175.6 | 175.8 | 176.4 | 177.4 | 179.5 |
| Q4_K_M | 174.28 GiB | 175.0 | 175.1 | 175.4 | 175.9 | 177.0 | 179.1 |
| I1-Q4_K_M | 173.97 GiB | 174.7 | 174.8 | 175.1 | 175.6 | 176.7 | 178.8 |
| Q4_K_S | 168.15 GiB | 168.9 | 169.0 | 169.3 | 169.8 | 170.9 | 173.0 |
| Q4_0 | 165.08 GiB | 165.8 | 165.9 | 166.2 | 166.7 | 167.8 | 169.9 |
| I1-Q4_K_S | 163.33 GiB | 164.1 | 164.2 | 164.5 | 165.0 | 166.0 | 168.2 |
| I1-Q4_0 | 162.70 GiB | 163.4 | 163.6 | 163.8 | 164.4 | 165.4 | 167.5 |
| IQ4_NL | 162.41 GiB | 163.1 | 163.3 | 163.5 | 164.1 | 165.1 | 167.2 |
| IQ4_XS | 153.50 GiB | 154.2 | 154.4 | 154.6 | 155.2 | 156.2 | 158.3 |
| I1-IQ4_XS | 153.13 GiB | 153.9 | 154.0 | 154.3 | 154.8 | 155.8 | 158.0 |
| I1-Q3_K_L | 148.65 GiB | 149.4 | 149.5 | 149.8 | 150.3 | 151.4 | 153.5 |
| Q3_K_M | 137.19 GiB | 137.9 | 138.0 | 138.3 | 138.8 | 139.9 | 142.0 |
| I1-Q3_K_M | 137.19 GiB | 137.9 | 138.0 | 138.3 | 138.8 | 139.9 | 142.0 |
| Q3_K_L | 136.05 GiB | 136.8 | 136.9 | 137.2 | 137.7 | 138.8 | 140.9 |
| IQ3_M | 130.65 GiB | 131.4 | 131.5 | 131.8 | 132.3 | 133.4 | 135.5 |
| I1-IQ3_M | 125.52 GiB | 126.3 | 126.4 | 126.6 | 127.2 | 128.2 | 130.3 |
| Q3_K_S | 124.51 GiB | 125.2 | 125.4 | 125.6 | 126.2 | 127.2 | 129.3 |
| I1-IQ3_S | 123.99 GiB | 124.7 | 124.8 | 125.1 | 125.6 | 126.7 | 128.8 |
| I1-Q3_K_S | 123.96 GiB | 124.7 | 124.8 | 125.1 | 125.6 | 126.7 | 128.8 |
| UD-IQ3_XXS | 118.70 GiB | 119.4 | 119.6 | 119.8 | 120.4 | 121.4 | 123.5 |
| IQ3_XS | 117.68 GiB | 118.4 | 118.5 | 118.8 | 119.3 | 120.4 | 122.5 |
| I1-IQ3_XS | 117.12 GiB | 117.8 | 118.0 | 118.2 | 118.8 | 119.8 | 121.9 |
| IQ3_XXS | 113.14 GiB | 113.9 | 114.0 | 114.3 | 114.8 | 115.8 | 118.0 |
| I1-IQ3_XXS | 110.41 GiB | 111.1 | 111.3 | 111.5 | 112.1 | 113.1 | 115.2 |
| Q2_K_L | 104.70 GiB | 105.4 | 105.6 | 105.8 | 106.4 | 107.4 | 109.5 |
| Q2_K | 104.57 GiB | 105.3 | 105.4 | 105.7 | 106.2 | 107.3 | 109.4 |
| I1-Q2_K | 104.57 GiB | 105.3 | 105.4 | 105.7 | 106.2 | 107.3 | 109.4 |
| UD-IQ2_M | 99.33 GiB | 100.1 | 100.2 | 100.5 | 101.0 | 102.0 | 104.1 |
| I1-Q2_K_S | 97.22 GiB | 98.0 | 98.1 | 98.3 | 98.9 | 99.9 | 102.0 |
| UD-IQ2_XXS | 94.58 GiB | 95.3 | 95.4 | 95.7 | 96.2 | 97.3 | 99.4 |
| I1-IQ2_M | 93.87 GiB | 94.6 | 94.7 | 95.0 | 95.5 | 96.6 | 98.7 |
| IQ2_M | 90.09 GiB | 90.8 | 91.0 | 91.2 | 91.7 | 92.8 | 94.9 |
| UD-IQ1_M | 87.36 GiB | 88.1 | 88.2 | 88.5 | 89.0 | 90.1 | 92.2 |
| I1-IQ2_S | 85.29 GiB | 86.0 | 86.2 | 86.4 | 86.9 | 88.0 | 90.1 |
| I1-IQ2_XS | 84.00 GiB | 84.7 | 84.9 | 85.1 | 85.7 | 86.7 | 88.8 |
| UD-IQ1_S | 81.23 GiB | 82.0 | 82.1 | 82.4 | 82.9 | 83.9 | 86.0 |
| IQ2_S | 79.34 GiB | 80.1 | 80.2 | 80.5 | 81.0 | 82.0 | 84.2 |
| IQ2_XS | 79.09 GiB | 79.8 | 80.0 | 80.2 | 80.7 | 81.8 | 83.9 |
| I1-IQ2_XXS | 75.37 GiB | 76.1 | 76.2 | 76.5 | 77.0 | 78.1 | 80.2 |
| UD-TQ1_0 | 72.36 GiB | 73.1 | 73.2 | 73.5 | 74.0 | 75.1 | 77.2 |
| IQ2_XXS | 68.47 GiB | 69.2 | 69.3 | 69.6 | 70.1 | 71.2 | 73.3 |
| I1-IQ1_M | 64.64 GiB | 65.4 | 65.5 | 65.8 | 66.3 | 67.3 | 69.5 |
| IQ1_M | 61.31 GiB | 62.0 | 62.2 | 62.4 | 63.0 | 64.0 | 66.1 |
| IQ1_S | 58.91 GiB | 59.6 | 59.8 | 60.0 | 60.6 | 61.6 | 63.7 |
| I1-IQ1_S | 58.20 GiB | 58.9 | 59.1 | 59.3 | 59.9 | 60.9 | 63.0 |
Figures are GiB of total memory: weights plus KV cache plus compute buffer and backend overhead. Weights and KV are near-exact; the overhead term is modeled. Hover any cell for the breakdown.
Why other calculators disagree
A parameters × bits ÷ 8 estimate ignores two things that dominate at long context. First, the weights themselves are not the nominal rate — quantizations are mixtures, so the real file is consistently larger than the label implies. Second, most of this model's layers cache only a 128-token window rather than the full context.