Magistral-Small-2507 - DASH-Q 2-bit GGUF
llama.cpp GGUF files of mistralai/Magistral-Small-2507 quantized with DASH-Q, at several 2-bit-class sizes. Every file uses only standard llama.cpp tensor types (no tensor above 4 bits) and loads in any recent llama.cpp build.
| File | Type | Size | Bits / weight |
|---|---|---|---|
Magistral-Small-2507-DASHQ-IQ2_XXS.gguf |
IQ2_XXS | 6.76 GB | 2.29 |
Magistral-Small-2507-DASHQ-IQ2_XS.gguf |
IQ2_XS | 7.68 GB | 2.60 |
Magistral-Small-2507-DASHQ-IQ2_M.gguf |
IQ2_M | 8.25 GB | 2.80 |
Magistral-Small-2507-DASHQ-Q2_K_XL.gguf |
Q2_K_XL | 9.15 GB | 3.10 |
Perplexity (lower is better)
| Type | Model | Size | WikiText-2 | C4 |
|---|---|---|---|---|
| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 6.55 GB | 33.28 | 75.00 |
| IQ2_XXS | unsloth UD-IQ2_XXS | 6.75 GB | 33.68 | 70.90 |
| IQ2_XXS | DASH-Q IQ2_XXS | 6.76 GB | 9.15 | 15.21 |
| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 7.21 GB | 22.50 | 47.15 |
| IQ2_XS | DASH-Q IQ2_XS | 7.68 GB | 7.93 | 13.16 |
| IQ2_M | llama.cpp IQ2_M (imatrix) | 8.11 GB | 13.67 | 24.86 |
| IQ2_M | unsloth UD-IQ2_M | 8.24 GB | 13.64 | 24.83 |
| IQ2_M | DASH-Q IQ2_M | 8.25 GB | 7.43 | 12.33 |
| Q2_K_XL | llama.cpp Q2_K (imatrix) | 8.89 GB | 9.74 | 15.46 |
| Q2_K_XL | unsloth UD-Q2_K_XL | 9.29 GB | 9.04 | 14.28 |
| Q2_K_XL | DASH-Q Q2_K_XL | 9.15 GB | 7.24 | 12.00 |
llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).
Usage
llama-cli -m Magistral-Small-2507-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192
License
Inherits the license of the base model (mistralai/Magistral-Small-2507).