Model comparison
TinyMistral-248M-v2-Instruct vs gemma-3-270m-it
At Q4_K_M, TinyMistral-248M-v2-Instruct is the smaller download — 155,668,768 bytes against 253,115,168.
From the file· summed bytes, KV per layer
Side by side
| TinyMistral-248M-v2-Instruct | gemma-3-270m-it | |
|---|---|---|
| Parameters | 248M | 268M |
| Architecture | llama | gemma3 |
| Layers | 12 | 18 |
| Native context | 32,768 | 32,768 |
| Mixture of experts | no | no |
| Quantizations published | 6 | 39 |
| Smallest quantization | 0.10 GiB | 0.17 GiB |
| Q4_K_M | 0.14 GiB | 0.24 GiB |
| Licence | apache-2.0 | gemma |
KV cache by context
the term that decides long-context viability
| Context | TinyMistral-248M-v2-Instruct | gemma-3-270m-it | Ratio |
|---|---|---|---|
| 4,096 | — | 0.03 GiB | — |
| 8,192 | — | 0.04 GiB | — |
| 16,384 | — | 0.06 GiB | — |
| 32,768 | — | 0.11 GiB | — |
| 65,536 | — | 0.20 GiB | — |
| 131,072 | — | 0.39 GiB | — |