Model comparison
Darwin-4B-Chimera vs HyperCLOVAX-SEED-Text-Instruct-1.5B
At Q4_K_M, HyperCLOVAX-SEED-Text-Instruct-1.5B is the smaller download — 1,133,974,368 bytes against 2,497,281,088.
From the file· summed bytes, KV per layer
Side by side
| Darwin-4B-Chimera | HyperCLOVAX-SEED-Text-Instruct-1.5B | |
|---|---|---|
| Parameters | 4.0B | 1.6B |
| Architecture | qwen3 | llama |
| Layers | 36 | — |
| Native context | 40,960 | — |
| Mixture of experts | no | no |
| Quantizations published | 59 | 1 |
| Smallest quantization | 0.98 GiB | 1.06 GiB |
| Q4_K_M | 2.33 GiB | 1.06 GiB |
| Licence | gemma | other |
KV cache by context
the term that decides long-context viability
| Context | Darwin-4B-Chimera | HyperCLOVAX-SEED-Text-Instruct-1.5B | Ratio |
|---|---|---|---|
| 4,096 | 0.56 GiB | — | — |
| 8,192 | 0.71 GiB | — | — |
| 16,384 | 0.90 GiB | — | — |
| 32,768 | 1.28 GiB | — | — |
| 65,536 | 2.03 GiB | — | — |
| 131,072 | 3.53 GiB | — | — |