Model comparison
Qwen3.8-27B vs Ling-mini-2.0
At Q4_K_M, Ling-mini-2.0 is the smaller download — 9,911,575,072 bytes against 16,810,714,336. At long context the gap widens: Ling-mini-2.0's KV cache at 32K is 1.6× smaller, which usually matters more than the difference in weights.
From the file· summed bytes, KV per layer
Side by side
| Qwen3.8-27B | Ling-mini-2.0 | |
|---|---|---|
| Parameters | 27.8B | 16.3B |
| Architecture | qwen35 | bailingmoe2 |
| Layers | 64 | 20 |
| Native context | 262,144 | 32,768 |
| Mixture of experts | no | yes, 256 experts |
| Quantizations published | 22 | 30 |
| Smallest quantization | 8.39 GiB | 4.34 GiB |
| Q4_K_M | 15.66 GiB | 9.23 GiB |
| Licence | apache-2.0 | — |
KV cache by context
the term that decides long-context viability
| Context | Qwen3.8-27B | Ling-mini-2.0 | Ratio |
|---|---|---|---|
| 4,096 | 0.25 GiB | 0.16 GiB | 1.60× |
| 8,192 | 0.50 GiB | 0.31 GiB | 1.60× |
| 16,384 | 1.00 GiB | 0.63 GiB | 1.60× |
| 32,768 | 2.00 GiB | 1.25 GiB | 1.60× |
| 65,536 | 4.00 GiB | 2.50 GiB | 1.60× |
| 131,072 | 8.00 GiB | 5.00 GiB | 1.60× |