Model comparison
Janus-Pro-7B vs Krea-2-Turbo
At Q4_K_M, Janus-Pro-7B is the smaller download — 4,223,359,936 bytes against 7,216,993,376.
From the file· summed bytes, KV per layer
Side by side
| Janus-Pro-7B | Krea-2-Turbo | |
|---|---|---|
| Parameters | 7.4B | 12.8B |
| Architecture | llama | qwen_image |
| Layers | 30 | — |
| Native context | 16,384 | — |
| Mixture of experts | no | no |
| Quantizations published | 36 | 33 |
| Smallest quantization | 1.61 GiB | 2.78 GiB |
| Q4_K_M | 3.93 GiB | 6.72 GiB |
| Licence | mit | other |
KV cache by context
the term that decides long-context viability
| Context | Janus-Pro-7B | Krea-2-Turbo | Ratio |
|---|---|---|---|
| 4,096 | 1.88 GiB | — | — |
| 8,192 | 3.75 GiB | — | — |
| 16,384 | 7.50 GiB | — | — |
| 32,768 | 15.00 GiB | — | — |
| 65,536 | 30.00 GiB | — | — |
| 131,072 | 60.00 GiB | — | — |