---

license: apache-2.0

base_model: mistralai/Devstral-Small-2-24B-Instruct-2512

base_model_relation: quantized

library_name: gguf

pipeline_tag: text-generation

tags:

• gguf
• llama.cpp
• dashq
• quantized

---

!DASH-Q

Devstral-Small-2-24B-Instruct-2512 - DASH-Q 2-bit GGUF

llama.cpp GGUF files of mistralai/Devstral-Small-2-24B-Instruct-2512 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 |

| --- | --- | --- | --- |

| Devstral-Small-2-24B-Instruct-2512-DASHQ-IQ2_XXS.gguf | IQ2_XXS | 6.76 GB | 2.29 |

| Devstral-Small-2-24B-Instruct-2512-DASHQ-IQ2_XS.gguf | IQ2_XS | 7.68 GB | 2.60 |

| Devstral-Small-2-24B-Instruct-2512-DASHQ-IQ2_M.gguf | IQ2_M | 8.25 GB | 2.80 |

| Devstral-Small-2-24B-Instruct-2512-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 | 15.98 | 34.21 |

| IQ2_XXS | unsloth UD-IQ2_XXS | 6.75 GB | 15.36 | 31.91 |

| IQ2_XXS | DASH-Q IQ2_XXS | 6.76 GB | 6.68 | 11.88 |

| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 7.21 GB | 11.54 | 22.39 |

| IQ2_XS | DASH-Q IQ2_XS | 7.68 GB | 5.96 | 10.63 |

| IQ2_M | llama.cpp IQ2_M (imatrix) | 8.11 GB | 7.83 | 14.30 |

| IQ2_M | unsloth UD-IQ2_M | 8.24 GB | 7.60 | 13.78 |

| IQ2_M | DASH-Q IQ2_M | 8.25 GB | 5.53 | 9.98 |

| Q2_K_XL | llama.cpp Q2_K (imatrix) | 8.89 GB | 6.12 | 10.80 |

| Q2_K_XL | unsloth UD-Q2_K_XL | 9.29 GB | 6.03 | 10.61 |

| Q2_K_XL | DASH-Q Q2_K_XL | 9.15 GB | 5.43 | 9.81 |

llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).

Usage

llama-cli -m Devstral-Small-2-24B-Instruct-2512-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192

License

Inherits the license of the base model (mistralai/Devstral-Small-2-24B-Instruct-2512).