Can I run cogito-671b-v2.1 on a Radeon RX 6500 XT?
Not at these settings. No indexed quantization of cogito-671b-v2.1 fits Radeon RX 6500 XT at any context we compute, with q8_0 KV. The smallest shipped quantization is 138.82 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◐ |
|---|---|---|---|---|---|---|---|
| BF16 | 1250.08 GiB | 1251.2 | 1251.3 | 1251.6 | 1252.2 | 1253.3 | 1255.6 |
| Q8_0 | 664.29 GiB | 665.4 | 665.6 | 665.8 | 666.4 | 667.6 | 669.8 |
| Q6_K | 513.45 GiB | 514.6 | 514.7 | 515.0 | 515.6 | 516.7 | 519.0 |
| Q5_K_M | 443.46 GiB | 444.6 | 444.7 | 445.0 | 445.6 | 446.7 | 449.0 |
| Q5_K_S | 430.86 GiB | 432.0 | 432.1 | 432.4 | 433.0 | 434.1 | 436.4 |
| Q4_1 | 391.87 GiB | 393.0 | 393.1 | 393.4 | 394.0 | 395.1 | 397.4 |
| Q4_K_M | 377.55 GiB | 378.7 | 378.8 | 379.1 | 379.7 | 380.8 | 383.1 |
| Q4_K_S | 354.89 GiB | 356.0 | 356.2 | 356.4 | 357.0 | 358.1 | 360.4 |
| Q4_0 | 353.99 GiB | 355.1 | 355.3 | 355.5 | 356.1 | 357.2 | 359.5 |
| IQ4_NL | 353.09 GiB | 354.2 | 354.4 | 354.6 | 355.2 | 356.4 | 358.6 |
| IQ4_XS | 333.73 GiB | 334.9 | 335.0 | 335.3 | 335.9 | 337.0 | 339.3 |
| Q3_K_M | 298.44 GiB | 299.6 | 299.7 | 300.0 | 300.6 | 301.7 | 304.0 |
| Q3_K_S | 270.50 GiB | 271.6 | 271.8 | 272.0 | 272.6 | 273.8 | 276.0 |
| UD-IQ3_XXS | 254.72 GiB | 255.8 | 256.0 | 256.3 | 256.8 | 258.0 | 260.3 |
| Q2_K_L | 229.02 GiB | 230.1 | 230.3 | 230.6 | 231.1 | 232.3 | 234.6 |
| Q2_K | 228.82 GiB | 229.9 | 230.1 | 230.4 | 230.9 | 232.1 | 234.4 |
| UD-IQ2_M | 212.81 GiB | 213.9 | 214.1 | 214.4 | 214.9 | 216.1 | 218.4 |
| UD-IQ2_XXS | 201.88 GiB | 203.0 | 203.1 | 203.4 | 204.0 | 205.1 | 207.4 |
| UD-IQ1_M | 187.19 GiB | 188.3 | 188.5 | 188.7 | 189.3 | 190.5 | 192.7 |
| UD-IQ1_S | 173.11 GiB | 174.2 | 174.4 | 174.7 | 175.2 | 176.4 | 178.6 |
| IQ2_XXS | 162.59 GiB | 163.7 | 163.9 | 164.1 | 164.7 | 165.8 | 168.1 |
| UD-TQ1_0 | 150.80 GiB | 151.9 | 152.1 | 152.3 | 152.9 | 154.1 | 156.3 |
| IQ1_M | 138.82 GiB | 139.9 | 140.1 | 140.4 | 140.9 | 142.1 | 144.4 |
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, this model uses latent attention and allocates no V cache at all, so any formula reading num_key_value_heads overstates its cache by more than an order of magnitude.