π Model Description
This repo contains GGUF .mmproj files intended for use in KoboldCpp alongside other GGUF models.
Please pick the correct projector model to load for your architecture (e.g. all Mistral 7B based models should use the Mistral 7B projector.)
Load them in KoboldCpp with --mmproj or from the "Model Files" tab
When enabled, make sure you refresh KoboldAI Lite, then click Add Img and upload your image. Clicking that image should show Vision is enabled.
π GGUF File List
| π Filename | π¦ Size | β‘ Download |
|---|---|---|
|
LLaMA3-8B_mmproj-Q4_1.gguf
LFS
Q4
|
187.73 MB | Download |
|
Ministral-3-3B-Instruct-2512-mmproj-Q8_0.gguf
LFS
Q8
|
426.37 MB | Download |
|
Qwen2-VL-2B-mmproj-q5_1.gguf
LFS
Q5
|
482.21 MB | Download |
|
Qwen2-VL-7B-mmproj-q5_1.gguf
LFS
Q5
|
489.72 MB | Download |
|
Qwen2.5-Omni-3B-mmproj-Q8_0.gguf
LFS
Q8
|
1.43 GB | Download |
|
Qwen2.5-Omni-7B-mmproj-Q8_0.gguf
LFS
Q8
|
1.44 GB | Download |
|
gemma3-12b-mmproj.gguf
LFS
|
814.63 MB | Download |
|
gemma3-27b-mmproj.gguf
LFS
|
818 MB | Download |
|
gemma3-4b-mmproj.gguf
LFS
|
811.82 MB | Download |
|
llama-13b-mmproj-v1.5.Q4_1.gguf
LFS
Q4
|
193.99 MB | Download |
|
llama-7b-mmproj-v1.5-Q4_0.gguf
Recommended
LFS
Q4
|
169.2 MB | Download |
|
minicpm-mmproj-model-f16.gguf
LFS
FP16
|
996.04 MB | Download |
|
mistral-7b-mmproj-v1.5-Q4_1.gguf
LFS
Q4
|
187.73 MB | Download |
|
mmproj-mistralai_Mistral-Small-3.1-24B-Instruct-2503-f16.gguf
LFS
FP16
|
837.38 MB | Download |
|
obsidian-3b_mmproj-Q4_1.gguf
LFS
Q4
|
180.69 MB | Download |
|
pixtral-12b-mmproj-f16.gguf
LFS
FP16
|
829.76 MB | Download |
|
qwen2.5-vl-7b-vision-mmproj-f16.gguf
LFS
FP16
|
1.26 GB | Download |
|
yi-34b-mmproj-v1.6-Q4_1.gguf
LFS
Q4
|
210.28 MB | Download |