Model comparison
TinyMistral-248M-v2-Instruct vs Qwen3-0.6B
At Q4_K_M, TinyMistral-248M-v2-Instruct is the smaller download — 155,668,768 bytes against 396,705,472.
From the file· summed bytes, KV per layer
Side by side
| TinyMistral-248M-v2-Instruct | Qwen3-0.6B | |
|---|---|---|
| Parameters | 248M | 752M |
| Architecture | llama | qwen3 |
| Layers | 12 | 28 |
| Native context | 32,768 | 40,960 |
| Mixture of experts | no | no |
| Quantizations published | 6 | 44 |
| Smallest quantization | 0.10 GiB | 0.20 GiB |
| Q4_K_M | 0.14 GiB | 0.37 GiB |
| Licence | apache-2.0 | — |
KV cache by context
the term that decides long-context viability
| Context | TinyMistral-248M-v2-Instruct | Qwen3-0.6B | 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 | — |