Mistral-Small-3.2-24B-Instruct-2506 - DASH-Q 2-bit GGUF
llama.cpp GGUF files of mistralai/Mistral-Small-3.2-24B-Instruct-2506 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 |
|---|---|---|---|
Mistral-Small-3.2-24B-Instruct-2506-DASHQ-IQ2_XXS.gguf |
IQ2_XXS | 6.76 GB | 2.29 |
Mistral-Small-3.2-24B-Instruct-2506-DASHQ-IQ2_XS.gguf |
IQ2_XS | 7.68 GB | 2.60 |
Mistral-Small-3.2-24B-Instruct-2506-DASHQ-IQ2_M.gguf |
IQ2_M | 8.25 GB | 2.80 |
Mistral-Small-3.2-24B-Instruct-2506-DASHQ-Q2_K_XL.gguf |
Q2_K_XL | 9.15 GB | 3.10 |
Text only: the vision tower is not included.
Perplexity (lower is better)
| Type | Model | Size | WikiText-2 | C4 |
|---|---|---|---|---|
| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 6.55 GB | 18.06 | 39.17 |
| IQ2_XXS | unsloth UD-IQ2_XXS | 6.75 GB | 17.23 | 35.00 |
| IQ2_XXS | DASH-Q IQ2_XXS | 6.76 GB | 6.63 | 11.48 |
| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 7.21 GB | 13.38 | 24.25 |
| IQ2_XS | DASH-Q IQ2_XS | 7.68 GB | 5.98 | 10.31 |
| IQ2_M | llama.cpp IQ2_M (imatrix) | 8.11 GB | 8.17 | 13.98 |
| IQ2_M | unsloth UD-IQ2_M | 8.24 GB | 7.74 | 13.26 |
| IQ2_M | DASH-Q IQ2_M | 8.25 GB | 5.59 | 9.73 |
| Q2_K_XL | llama.cpp Q2_K (imatrix) | 8.89 GB | 6.26 | 10.48 |
| Q2_K_XL | unsloth UD-Q2_K_XL | 9.29 GB | 5.99 | 10.10 |
| Q2_K_XL | DASH-Q Q2_K_XL | 9.15 GB | 5.47 | 9.54 |
llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).
Usage
llama-cli -m Mistral-Small-3.2-24B-Instruct-2506-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192
License
Inherits the license of the base model (mistralai/Mistral-Small-3.2-24B-Instruct-2506).