Model comparison
Olmo-3.1-32B-Instruct vs Qwen3-30B-A3B-Thinking-2507
At Q4_K_M, Qwen3-30B-A3B-Thinking-2507 is the smaller download — 18,556,685,824 bytes against 19,482,034,976.
From the file· summed bytes, KV per layer
Side by side
| Olmo-3.1-32B-Instruct | Qwen3-30B-A3B-Thinking-2507 | |
|---|---|---|
| Parameters | 32.2B | 30.5B |
| Architecture | olmo2 | qwen3moe |
| Layers | 64 | 48 |
| Native context | 65,536 | 262,144 |
| Mixture of experts | no | yes, 128 experts |
| Quantizations published | 49 | 51 |
| Smallest quantization | 6.75 GiB | 7.05 GiB |
| Q4_K_M | 18.14 GiB | 17.28 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 | Qwen3-30B-A3B-Thinking-2507 | Ratio |
|---|---|---|---|
| 4,096 | 1.00 GiB | 0.38 GiB | 2.67× |
| 8,192 | 1.34 GiB | 0.75 GiB | 1.79× |
| 16,384 | 1.84 GiB | 1.50 GiB | 1.23× |
| 32,768 | 2.84 GiB | 3.00 GiB | 1.05× |
| 65,536 | 4.84 GiB | 6.00 GiB | 1.24× |
| 131,072 | 8.84 GiB | 12.00 GiB | 1.36× |