Can I run gemma-4-E4B-it on a Radeon RX 6400?
Not at these settings. No indexed quantization of gemma-4-E4B-it fits Radeon RX 6400 at any context we compute, with q4_0 KV. The smallest shipped quantization is 3.30 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 | 14.18 GiB | 15.1 | 15.1 | 15.2 | 15.2 | 15.4 | 15.6 |
| Q8_0 | 7.63 GiB | 8.6 | 8.6 | 8.6 | 8.7 | 8.8 | 9.1 |
| Q6_K | 6.59 GiB | 7.5 | 7.6 | 7.6 | 7.6 | 7.8 | 8.0 |
| Q5_K_M | 5.37 GiB | 6.3 | 6.3 | 6.4 | 6.4 | 6.5 | 6.8 |
| Q5_K_S | 5.03 GiB | 6.0 | 6.0 | 6.0 | 6.1 | 6.2 | 6.5 |
| Q4_K_M | 4.97 GiB | 5.9 | 5.9 | 6.0 | 6.0 | 6.2 | 6.4 |
| Q4_1 | 4.73 GiB | 5.7 | 5.7 | 5.7 | 5.8 | 5.9 | 6.2 |
| IQ4_XS | 4.72 GiB | 5.7 | 5.7 | 5.7 | 5.8 | 5.9 | 6.2 |
| Q3_K_M | 4.52 GiB | 5.5 | 5.5 | 5.5 | 5.6 | 5.7 | 5.9 |
| Q4_K_S | 4.51 GiB | 5.5 | 5.5 | 5.5 | 5.6 | 5.7 | 5.9 |
| Q4_0 | 4.50 GiB | 5.5 | 5.5 | 5.5 | 5.6 | 5.7 | 5.9 |
| IQ4_NL | 4.50 GiB | 5.5 | 5.5 | 5.5 | 5.6 | 5.7 | 5.9 |
| IQ3_M | 4.39 GiB | 5.3 | 5.4 | 5.4 | 5.4 | 5.6 | 5.8 |
| Q3_K_S | 3.60 GiB | 4.5 | 4.6 | 4.6 | 4.7 | 4.8 | 5.0 |
| UD-IQ3_XXS | 3.46 GiB | 4.4 | 4.4 | 4.5 | 4.5 | 4.6 | 4.9 |
| UD-IQ2_M | 3.30 GiB | 4.3 | 4.3 | 4.3 | 4.4 | 4.5 | 4.7 |
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 512-token window rather than the full context.