Skip to content

All on-device models

Zero-shot image classification models for iPhone and Mac (Core ML & MLX)

21 models for this task are tagged Core ML or MLX on Hugging Face, or come from Apple: 8 Core ML, 2 MLX and 11 other Apple models. The table lists the most downloaded in the last 30 days.

Last updated 2026-11-02 (UTC).

Showing 21 of 21.

ModelSizeDownloads (30 days)
apple/MobileCLIP-S2-OpenCLIPApple, other format800 MB32K
apple/MobileCLIP-S1-OpenCLIPApple, other format685 MB12.3K
apple/aimv2-large-patch14-224-litMLX5.2 GB8,133
apple/MobileCLIP-B-LT-OpenCLIPApple, other format1.2 GB5,947
apple/MobileCLIP-B-OpenCLIPApple, other format—1,182
mlx-community/siglip2-so400m-patch16-384MLX—1,152
apple/TiC-CLIP-basic-cumulativeApple, other format—397
apple/TiC-CLIP-bestpool-sequentialApple, other format—379
apple/TiC-CLIP-bestpool-cumulativeApple, other format—359
apple/TiC-CLIP-basic-sequentialApple, other format—350
FluidInference/siglip2-base-patch16-256-coremlCore MLfrom google/siglip2-base-patch16-256—294
ddtfzco/decluttr-clip-vit-b32-coremlCore MLfrom openai/clip-vit-base-patch32—215
OZU-Technology/CinemaCLIPCore MLfrom laion/CLIP-ViT-B-32-256x256-DataComp-s34B-b86K—155
apple/DFN-publicApple, other format—110
pablete/CinemaCLIPCore MLfrom laion/CLIP-ViT-B-32-256x256-DataComp-s34B-b86K—43
batmac/ViT-B-16-SigLIP2-Image-CoreMLCore MLfrom google/siglip2-base-patch16-224—36
apple/TiC-CLIP-basic-oracleApple, other format—28
apple/TiC-CLIP-bestpool-oracleApple, other format—27
Crjstobal/siglip2-base-patch16-256-coremlCore MLfrom google/siglip2-base-patch16-256—23
f-loren/fototag-siglip2-coremlCore MLfrom google/siglip2-base-patch16-256—19
LucidexGroup/threadfinder-siglip2-base-patch16-224Core ML—17

Reading the table

  • Downloads are Hugging Face's count for the last 30 days, not all time. Hugging Face counts requests for a model's main file, so one install can count more than once.
  • Size is the total of the repo's current files on Hugging Face, in the sizes it shows (1 GB = 1,000 MB). Some repos hold several variants of a model, so it can be larger than the one download you need; a dash means it has not been measured yet.
  • Trending is Hugging Face's own score for recent attention. "from" under a name is the model it was made from.
  • Parameters is the parameter count in the model files. A count marked * comes from the model's name (such as 27B) instead: the files gave none, or a much smaller one, as packed low-bit files can.
  • Quantization is read from the model's tags and name, not checked against the files.

Before you use one

The list comes from Hugging Face's public API and is updated daily. We don't test each model. Check the license and read the model card before you ship one in an app: some licenses limit commercial use, and "other" means the terms are on the model card. Models whose names or tags point at adult or safety-removed content are left out, and so are gated models that need a Hugging Face sign-in to download.

Other tasks