v1.0.11.0 is here: six tools, one workspace.See what's new →
Armor EdgeAI
EN
Download

AI Models

AI Models

Armor EdgeAI's features use the AI models listed below. The installer contains no models: during first-time setup the App downloads the models your chosen features need from our download service, and other models are downloaded only when you choose them. All of them run on your computer.

Each model is provided under its own license; select a license name to read the full terms. "Original model" is the version published by its authors. "Obtained from" is where the file this product actually uses comes from — usually a quantized or converted build of the original.

40Models
4License types
v0012Model list version
ModelsUsed forLicenseSource
Qwen 3.5 0.8BAlibaba QwenGGUF Q4_K_M · 533 MB + 205 MB vision componentChat · understands imagesApache-2.0Original model Qwen/Qwen3.5-0.8BObtained from unsloth/Qwen3.5-0.8B-GGUF
Qwen 3.5 2B VisionAlibaba QwenGGUF Q4_K_M · 1.3 GB + 668 MB vision componentChat · understands imagesApache-2.0Original model Qwen/Qwen3.5-2BObtained from unsloth/Qwen3.5-2B-GGUF
Qwen 3.5 4B VisionAlibaba QwenGGUF UD-Q4_K_XL · 2.9 GB + 672 MB vision componentChat · understands imagesApache-2.0Original model Qwen/Qwen3.5-4BObtained from unsloth/Qwen3.5-4B-GGUF
Qwen 3.5 9BAlibaba QwenGGUF Q4_K_M · 5.7 GBChatApache-2.0Original model Qwen/Qwen3.5-9BObtained from unsloth/Qwen3.5-9B-GGUF
Qwen 3.6 27B VisionAlibaba QwenGGUF UD-Q4_K_XL · 17.6 GB + 928 MB vision componentChat · understands imagesApache-2.0Original model Qwen/Qwen3.6-27BObtained from unsloth/Qwen3.6-27B-GGUF
Qwen 3.6 35B-A3BAlibaba QwenGGUF UD-Q4_K_M · 22.1 GB + 899 MB vision componentChat · understands imagesApache-2.0Original model Qwen/Qwen3.6-35B-A3BObtained from unsloth/Qwen3.6-35B-A3B-GGUF
Qwen 3.6 35B-A3B(MTP)Alibaba QwenGGUF UD-Q4_K_M · 22.7 GB + 899 MB vision componentChat · understands imagesApache-2.0Original model Qwen/Qwen3.6-35B-A3BObtained from unsloth/Qwen3.6-35B-A3B-MTP-GGUF
Gemma 4 E2BGoogleGGUF Q4_K_M · 3.1 GB + 986 MB vision componentChat · understands imagesApache-2.0Original model google/gemma-4-E2B-itObtained from unsloth/gemma-4-E2B-it-GGUF
Gemma 4 E4BGoogleGGUF Q4_K_M · 5.0 GB + 990 MB vision componentChat · understands imagesApache-2.0Original model google/gemma-4-E4B-itObtained from unsloth/gemma-4-E4B-it-GGUF
Gemma 4 26B-A4BGoogleGGUF UD-Q4_K_M · 16.9 GB + 1.2 GB vision componentChat · understands imagesApache-2.0Original model google/gemma-4-26B-A4B-itObtained from unsloth/gemma-4-26B-A4B-it-GGUF
Gemma 4 31BGoogleGGUF Q4_K_M · 18.3 GB + 1.2 GB vision componentChat · understands imagesApache-2.0Original model google/gemma-4-31B-itObtained from unsloth/gemma-4-31B-it-GGUF
Ministral 3 3BMistral AIGGUF Q4_K_M · 2.1 GB + 840 MB vision componentChat · understands imagesApache-2.0Original model mistralai/Ministral-3-3B-Instruct-2512Obtained from unsloth/Ministral-3-3B-Instruct-2512-GGUF
Ministral 3 8BMistral AIGGUF Q4_K_M · 5.2 GB + 857 MB vision componentChat · understands imagesApache-2.0Original model mistralai/Ministral-3-8B-Instruct-2512Obtained from unsloth/Ministral-3-8B-Instruct-2512-GGUF
Ministral 3 14BMistral AIGGUF Q4_K_M · 8.2 GB + 878 MB vision componentChat · understands imagesApache-2.0Original model mistralai/Ministral-3-14B-Instruct-2512Obtained from unsloth/Ministral-3-14B-Instruct-2512-GGUF
gpt-oss 20BOpenAIGGUF Q4_K_M · 11.6 GBChatApache-2.0Original model openai/gpt-oss-20bObtained from unsloth/gpt-oss-20b-GGUF
Muse Glimmer 30BMeta Superintelligence LabGGUF UD-Q4_K_XL · 15.9 GB + 2.1 GB vision componentChat · understands imagesApache-2.0Original model meta-models/Muse-Glimmer-30BObtained from unsloth/Muse-Glimmer-30B-GGUF
Qwen 3.8 27B VisionAlibaba QwenGGUF UD-Q4_K_M · 16.5 GB + 928 MB vision componentChat · understands imagesApache-2.0Original model Qwen/Qwen3.8-27BObtained from unsloth/Qwen3.8-27B-GGUF
MiMo V2.6 Distill 9BXiaomi MiMoGGUF Q4_K_M · 5.8 GB + 918 MB vision componentChat · understands imagesMITThe post-trained weights are MIT; the base model Qwen3.5-9B is Apache-2.0. Both licenses apply.Original model XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9BQuantized by us from the original weights
Z-Image TurboAlibaba Tongyi-MAIGGUF Q4_K_M · 5.0 GBImage generationApache-2.0Original model Tongyi-MAI/Z-Image-TurboObtained from unsloth/Z-Image-Turbo-GGUF
Z-Image VAEsafetensors · 335 MBImage decoder (VAE) for Z-ImageApache-2.0Obtained from Comfy-Org/z_image_turbo
Qwen3 4B Instruct 2507Alibaba QwenGGUF Q4_K_M · 2.5 GBText encoder for Z-ImageApache-2.0Original model Qwen/Qwen3-4B-Instruct-2507Obtained from unsloth/Qwen3-4B-Instruct-2507-GGUF
FLUX.2 klein Base 4BBlack Forest LabsGGUF Q4_K_M · 2.6 GBImage generationApache-2.0Original model black-forest-labs/FLUX.2-klein-base-4BObtained from unsloth/FLUX.2-klein-base-4B-GGUF
Qwen3 4BAlibaba QwenGGUF Q4_K_M · 2.5 GBText encoder for FLUX.2Apache-2.0Original model Qwen/Qwen3-4BObtained from unsloth/Qwen3-4B-GGUF
FLUX.2 VAEsafetensors · 336 MBImage decoder (VAE) for FLUX.2Apache-2.0Obtained from Comfy-Org/ERNIE-Image
Qwen-Image 2512Alibaba QwenGGUF Q4_K_M · 13.2 GBImage generationApache-2.0Original model Qwen/Qwen-Image-2512Obtained from unsloth/Qwen-Image-2512-GGUF
Qwen2.5-VL 7B InstructAlibaba QwenGGUF UD-Q4_K_XL · 4.8 GBText encoder for Qwen-ImageApache-2.0Original model Qwen/Qwen2.5-VL-7B-InstructObtained from unsloth/Qwen2.5-VL-7B-Instruct-GGUF
Qwen-Image VAEAlibaba Qwensafetensors · 254 MBImage decoder (VAE) for Qwen-ImageApache-2.0Original model Qwen/Qwen-ImageObtained from Comfy-Org/Qwen-Image_ComfyUI
Real-ESRGAN x4plusXintao Wang (Tencent ARC Lab)PyTorch (.pth) · 67 MBImage enhancement (4× upscaling)BSD-3-ClauseObtained from xinntao/Real-ESRGAN
PaddleOCR-VL 1.6Baidu PaddlePaddleGGUF · 936 MB + 882 MB vision componentDocument text recognitionApache-2.0Obtained from PaddlePaddle/PaddleOCR-VL-1.6-GGUF
PP-DocLayoutV3Baidu PaddlePaddleONNX · 131 MBDocument layout detectionApache-2.0Original model PaddlePaddle/PP-DocLayoutV3Obtained from alex-dinh/PP-DocLayoutV3-ONNX
Whisper SmallOpenAIggml q5_1 · 190 MBSpeech to textMITOriginal model openai/whisperObtained from ggerganov/whisper.cpp
Whisper Large v3 TurboOpenAIggml q5_0 · 574 MBSpeech to textMITOriginal model openai/whisperObtained from ggerganov/whisper.cpp
Whisper Large v3OpenAIggml q5_0 · 1.1 GBSpeech to textMITOriginal model openai/whisperObtained from ggerganov/whisper.cpp
Silero VAD v6.2Sileroggml · 885 KBVoice activity detectionMITOriginal model snakers4/silero-vadObtained from ggml-org/whisper-vad
Nemotron 3 DiarizationNVIDIAGGUF q8_0 · 107 MBSpeaker diarizationOpenMDW-1.1Obtained from nvidia/Nemotron-3-Diarization
Nemotron 3.5 ASR Streaming 0.6BNVIDIAGGUF q8_0 · 742 MBLive captionsOpenMDW-1.1Obtained from nvidia/nemotron-3.5-asr-streaming-0.6b
YuNetShiqi Yu et al.ONNX · 233 KBFace detectionMITOriginal model ShiqiYu/libfacedetection.trainObtained from opencv/opencv_zoo
SFaceYaoyao ZhongONNX · 39 MBFace recognitionApache-2.0Original model zhongyy/SFaceObtained from opencv/opencv_zoo
AuraFace v1falONNX · 261 MBFace recognition (high accuracy)Apache-2.0Obtained from fal/AuraFace-v1
Qwen3 Embedding 0.6BAlibaba QwenGGUF Q8_0 · 639 MBSemantic searchApache-2.0Obtained from Qwen/Qwen3-Embedding-0.6B-GGUF

Showing 40 models · from model catalog v0012 (published 2026-10-23).

For the licenses of the inference engines and other software, see Third-Party Licenses.