Model comparison
Mistral-Large-3-675B-Instruct-2512 vs DeepSeek-V4-Flash
These two publish different quantization sets; the table below has the exact sizes.
From the file· summed bytes, KV per layer
Side by side
| Mistral-Large-3-675B-Instruct-2512 | DeepSeek-V4-Flash | |
|---|---|---|
| Parameters | 673B | 291B |
| Architecture | deepseek2 | deepseek4 |
| Layers | — | 43 |
| Native context | — | 1,048,576 |
| Mixture of experts | no | yes, 256 experts |
| Quantizations published | 45 | 12 |
| Smallest quantization | 132.50 GiB | 76.87 GiB |
| Q4_K_M | 379.04 GiB | — |
| Licence | apache-2.0 | mit |
KV cache by context
the term that decides long-context viability
| Context | Mistral-Large-3-675B-Instruct-2512 | DeepSeek-V4-Flash | Ratio |
|---|---|---|---|
| 4,096 | — | 0.06 GiB | — |
| 8,192 | — | 0.06 GiB | — |
| 16,384 | — | 0.06 GiB | — |
| 32,768 | — | 0.06 GiB | — |
| 65,536 | — | 0.06 GiB | — |
| 131,072 | — | 0.06 GiB | — |