Model comparison
Atomight-V2.5-1.7B vs Wan2.1-T2V-1.3B
At Q4_K_M, Wan2.1-T2V-1.3B is the smaller download — 982,716,640 bytes against 1,107,410,112.
From the file· summed bytes, KV per layer
Side by side
| Atomight-V2.5-1.7B | Wan2.1-T2V-1.3B | |
|---|---|---|
| Parameters | 1.7B | 1.4B |
| Architecture | qwen3 | qwen35 |
| Layers | 28 | — |
| Native context | 40,960 | — |
| Mixture of experts | no | no |
| Quantizations published | 36 | 31 |
| Smallest quantization | 0.48 GiB | 0.61 GiB |
| Q4_K_M | 1.03 GiB | 0.92 GiB |
| Licence | mit | — |
KV cache by context
the term that decides long-context viability
| Context | Atomight-V2.5-1.7B | Wan2.1-T2V-1.3B | Ratio |
|---|---|---|---|
| 4,096 | 0.44 GiB | — | — |
| 8,192 | 0.88 GiB | — | — |
| 16,384 | 1.75 GiB | — | — |
| 32,768 | 3.50 GiB | — | — |
| 65,536 | 7.00 GiB | — | — |
| 131,072 | 14.00 GiB | — | — |