Model comparison
Qwen-Image-Edit-Rapid-AIO vs Qwen3.6-27B
At Q4_K_M, Qwen-Image-Edit-Rapid-AIO is the smaller download — 4,683,073,952 bytes against 16,547,397,888.
From the file· summed bytes, KV per layer
Side by side
| Qwen-Image-Edit-Rapid-AIO | Qwen3.6-27B | |
|---|---|---|
| Parameters | 20.4B | 27.8B |
| Architecture | qwen_image | qwen35 |
| Layers | — | 64 |
| Native context | — | 262,144 |
| Mixture of experts | no | no |
| Quantizations published | 72 | 40 |
| Smallest quantization | 4.36 GiB | 8.74 GiB |
| Q4_K_M | 4.36 GiB | 15.41 GiB |
| Licence | — | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | Qwen-Image-Edit-Rapid-AIO | Qwen3.6-27B | Ratio |
|---|---|---|---|
| 4,096 | — | 0.25 GiB | — |
| 8,192 | — | 0.50 GiB | — |
| 16,384 | — | 1.00 GiB | — |
| 32,768 | — | 2.00 GiB | — |
| 65,536 | — | 4.00 GiB | — |
| 131,072 | — | 8.00 GiB | — |