Model comparison
MN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-Thinking vs Qwen3.6-27B
At Q4_K_M, MN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-Thinking is the smaller download — 14,354,741,376 bytes against 16,547,397,888. At long context the gap widens: Qwen3.6-27B's KV cache at 32K is 5.1× smaller, which usually matters more than the difference in weights.
From the file· summed bytes, KV per layer
Side by side
| MN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-Thinking | Qwen3.6-27B | |
|---|---|---|
| Parameters | 23.4B | 27.8B |
| Architecture | llama | qwen35 |
| Layers | 81 | 64 |
| Native context | 1,024,000 | 262,144 |
| Mixture of experts | no | no |
| Quantizations published | 33 | 40 |
| Smallest quantization | 5.00 GiB | 8.74 GiB |
| Q4_K_M | 13.37 GiB | 15.41 GiB |
| Licence | apache-2.0 | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | MN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-Thinking | Qwen3.6-27B | Ratio |
|---|---|---|---|
| 4,096 | 1.27 GiB | 0.25 GiB | 5.06× |
| 8,192 | 2.53 GiB | 0.50 GiB | 5.06× |
| 16,384 | 5.06 GiB | 1.00 GiB | 5.06× |
| 32,768 | 10.13 GiB | 2.00 GiB | 5.06× |
| 65,536 | 20.25 GiB | 4.00 GiB | 5.06× |
| 131,072 | 40.50 GiB | 8.00 GiB | 5.06× |