Olmo-3.1-32B-Instruct - DASH-Q 2-bit GGUF
llama.cpp GGUF files of allenai/Olmo-3.1-32B-Instruct 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 |
|---|---|---|---|
Olmo-3.1-32B-Instruct-DASHQ-IQ2_XXS.gguf |
IQ2_XXS | 8.92 GB | 2.21 |
Olmo-3.1-32B-Instruct-DASHQ-IQ2_XS.gguf |
IQ2_XS | 10.22 GB | 2.53 |
Olmo-3.1-32B-Instruct-DASHQ-IQ2_M.gguf |
IQ2_M | 11.05 GB | 2.74 |
Olmo-3.1-32B-Instruct-DASHQ-Q2_K_XL.gguf |
Q2_K_XL | 12.22 GB | 3.03 |
Perplexity (lower is better)
| Type | Model | Size | WikiText-2 | C4 |
|---|---|---|---|---|
| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 8.76 GB | 8.33 | 15.44 |
| IQ2_XXS | unsloth UD-IQ2_XXS | 8.91 GB | 8.21 | 15.24 |
| IQ2_XXS | DASH-Q IQ2_XXS | 8.92 GB | 7.88 | 15.14 |
| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 9.69 GB | 7.91 | 15.02 |
| IQ2_XS | DASH-Q IQ2_XS | 10.22 GB | 7.28 | 14.22 |
| IQ2_M | llama.cpp IQ2_M (imatrix) | 10.97 GB | 7.43 | 14.43 |
| IQ2_M | unsloth UD-IQ2_M | 11.05 GB | 7.35 | 14.25 |
| IQ2_M | DASH-Q IQ2_M | 11.05 GB | 7.06 | 14.00 |
| Q2_K_XL | llama.cpp Q2_K (imatrix) | 12.01 GB | 7.25 | 14.18 |
| Q2_K_XL | unsloth UD-Q2_K_XL | 12.33 GB | 7.11 | 13.89 |
| Q2_K_XL | DASH-Q Q2_K_XL | 12.22 GB | 6.94 | 13.77 |
llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).
Usage
llama-cli -m Olmo-3.1-32B-Instruct-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192
License
Inherits the license of the base model (allenai/Olmo-3.1-32B-Instruct).