Can I run command-a-plus-05-2026-bf16 on a Radeon RX 6500 XT?
Not at these settings. No indexed quantization of command-a-plus-05-2026-bf16 fits Radeon RX 6500 XT at any context we compute, with q4_0 KV. The smallest shipped quantization is 45.87 GiB in weights alone, against 3.72 GiB usable. CPU offload can still run it, slowly.
Every quantization at every context
| Quant | Weights● | 4K◐ | 8K◐ | 16K◐ | 32K◐ | 64K◐ | 128K◐ |
|---|---|---|---|---|---|---|---|
| Q8_0 | 216.03 GiB | 217.1 | 217.1 | 217.2 | 217.3 | 217.6 | 218.2 |
| Q6_K | 176.79 GiB | 177.8 | 177.9 | 178.0 | 178.1 | 178.4 | 178.9 |
| Q5_K_M | 146.93 GiB | 148.0 | 148.0 | 148.1 | 148.2 | 148.5 | 149.1 |
| Q5_K_S | 141.69 GiB | 142.7 | 142.8 | 142.9 | 143.0 | 143.3 | 143.8 |
| Q4_1 | 129.29 GiB | 130.3 | 130.4 | 130.5 | 130.6 | 130.9 | 131.4 |
| Q4_K_L | 126.06 GiB | 127.1 | 127.2 | 127.2 | 127.4 | 127.6 | 128.2 |
| Q4_K_M | 125.81 GiB | 126.9 | 126.9 | 127.0 | 127.1 | 127.4 | 128.0 |
| Q4_K_S | 120.92 GiB | 122.0 | 122.0 | 122.1 | 122.2 | 122.5 | 123.1 |
| Q4_0 | 117.42 GiB | 118.5 | 118.5 | 118.6 | 118.7 | 119.0 | 119.6 |
| IQ4_NL | 116.79 GiB | 117.8 | 117.9 | 118.0 | 118.1 | 118.4 | 118.9 |
| IQ4_XS | 110.67 GiB | 111.7 | 111.8 | 111.8 | 112.0 | 112.3 | 112.8 |
| IQ3_M | 99.41 GiB | 100.5 | 100.5 | 100.6 | 100.7 | 101.0 | 101.6 |
| Q3_K_L | 99.27 GiB | 100.3 | 100.4 | 100.4 | 100.6 | 100.9 | 101.4 |
| Q3_K_M | 95.53 GiB | 96.6 | 96.6 | 96.7 | 96.8 | 97.1 | 97.7 |
| IQ3_XS | 95.16 GiB | 96.2 | 96.3 | 96.3 | 96.5 | 96.7 | 97.3 |
| Q3_K_S | 91.28 GiB | 92.3 | 92.4 | 92.4 | 92.6 | 92.9 | 93.4 |
| IQ3_XXS | 87.48 GiB | 88.5 | 88.6 | 88.6 | 88.8 | 89.1 | 89.6 |
| Q2_K_L | 74.80 GiB | 75.8 | 75.9 | 76.0 | 76.1 | 76.4 | 76.9 |
| Q2_K | 74.56 GiB | 75.6 | 75.7 | 75.7 | 75.9 | 76.1 | 76.7 |
| IQ2_M | 71.32 GiB | 72.4 | 72.4 | 72.5 | 72.6 | 72.9 | 73.5 |
| IQ2_S | 65.10 GiB | 66.1 | 66.2 | 66.3 | 66.4 | 66.7 | 67.3 |
| IQ2_XS | 63.98 GiB | 65.0 | 65.1 | 65.1 | 65.3 | 65.6 | 66.1 |
| IQ2_XXS | 57.98 GiB | 59.0 | 59.1 | 59.1 | 59.3 | 59.6 | 60.1 |
| IQ1_M | 50.59 GiB | 51.6 | 51.7 | 51.8 | 51.9 | 52.2 | 52.7 |
| IQ1_S | 45.87 GiB | 46.9 | 47.0 | 47.0 | 47.2 | 47.5 | 48.0 |
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 4,096-token window rather than the full context.