Model comparison
Muse-Glimmer-30B vs Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTP
At Q4_K_M, Muse-Glimmer-30B is the smaller download — 17,306,324,000 bytes against 36,545,830,912. At long context the gap widens: Muse-Glimmer-30B's KV cache at 32K is 4.0× smaller, which usually matters more than the difference in weights.
From the file· summed bytes, KV per layer
Side by side
| Muse-Glimmer-30B | Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTP | |
|---|---|---|
| Parameters | 29.8B | 27.8B |
| Architecture | muse-glimmer | qwen35 |
| Layers | 52 | 64 |
| Native context | 131,072 | 262,144 |
| Mixture of experts | no | no |
| Quantizations published | 36 | 34 |
| Smallest quantization | 8.31 GiB | 10.12 GiB |
| Q4_K_M | 16.12 GiB | 34.04 GiB |
| Licence | apache-2.0 | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | Muse-Glimmer-30B | Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTP | Ratio |
|---|---|---|---|
| 4,096 | 0.15 GiB | 0.25 GiB | 1.71× |
| 8,192 | 0.20 GiB | 0.50 GiB | 2.54× |
| 16,384 | 0.30 GiB | 1.00 GiB | 3.35× |
| 32,768 | 0.50 GiB | 2.00 GiB | 3.99× |
| 65,536 | 0.91 GiB | 4.00 GiB | 4.41× |
| 131,072 | 1.72 GiB | 8.00 GiB | 4.65× |