Model comparison
Olmo-3.1-32B-Instruct vs Qwen3.8-27B
At Q4_K_M, Qwen3.8-27B is the smaller download — 16,810,714,336 bytes against 19,482,034,976. At long context the gap widens: Qwen3.8-27B's KV cache at 32K is 1.4× smaller, which usually matters more than the difference in weights.
From the file· summed bytes, KV per layer
Side by side
| Olmo-3.1-32B-Instruct | Qwen3.8-27B | |
|---|---|---|
| Parameters | 32.2B | 27.8B |
| Architecture | olmo2 | qwen35 |
| Layers | 64 | 64 |
| Native context | 65,536 | 262,144 |
| Mixture of experts | no | no |
| Quantizations published | 49 | 22 |
| Smallest quantization | 6.75 GiB | 8.39 GiB |
| Q4_K_M | 18.14 GiB | 15.66 GiB |
| Licence | apache-2.0 | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | Olmo-3.1-32B-Instruct | Qwen3.8-27B | Ratio |
|---|---|---|---|
| 4,096 | 1.00 GiB | 0.25 GiB | 4.00× |
| 8,192 | 1.34 GiB | 0.50 GiB | 2.69× |
| 16,384 | 1.84 GiB | 1.00 GiB | 1.84× |
| 32,768 | 2.84 GiB | 2.00 GiB | 1.42× |
| 65,536 | 4.84 GiB | 4.00 GiB | 1.21× |
| 131,072 | 8.84 GiB | 8.00 GiB | 1.11× |