Model comparison
Nemotron-Cascade-2-30B-A3B vs Qwen3.8-27B
At Q4_K_M, Qwen3.8-27B is the smaller download — 16,810,714,336 bytes against 24,725,740,992. At long context the gap widens: Nemotron-Cascade-2-30B-A3B's KV cache at 32K is 1.2× smaller, which usually matters more than the difference in weights.
From the file· summed bytes, KV per layer
Side by side
| Nemotron-Cascade-2-30B-A3B | Qwen3.8-27B | |
|---|---|---|
| Parameters | 31.6B | 27.8B |
| Architecture | nemotron_h_moe | qwen35 |
| Layers | 52 | 64 |
| Native context | 262,144 | 262,144 |
| Mixture of experts | yes, 128 experts | no |
| Quantizations published | 49 | 22 |
| Smallest quantization | 16.64 GiB | 8.39 GiB |
| Q4_K_M | 23.03 GiB | 15.66 GiB |
| Licence | other | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | Nemotron-Cascade-2-30B-A3B | Qwen3.8-27B | Ratio |
|---|---|---|---|
| 4,096 | 0.20 GiB | 0.25 GiB | 1.23× |
| 8,192 | 0.41 GiB | 0.50 GiB | 1.23× |
| 16,384 | 0.81 GiB | 1.00 GiB | 1.23× |
| 32,768 | 1.63 GiB | 2.00 GiB | 1.23× |
| 65,536 | 3.25 GiB | 4.00 GiB | 1.23× |
| 131,072 | 6.50 GiB | 8.00 GiB | 1.23× |