Skip to content

All on-device models

Embedding models for iPhone and Mac (Core ML & MLX)

48 models for this task are tagged Core ML or MLX on Hugging Face, or come from Apple: 30 Core ML, 18 MLX and 0 other Apple models. The table lists the most downloaded in the last 30 days.

Last updated 2026-11-02 (UTC).

Showing 48 of 48.

ModelSizeDownloads (30 days)
jinaai/jina-embeddings-v2-small-enCore ML523 MB1.3M
jinaai/jina-embeddings-v2-base-enCore ML2.2 GB357K
mlx-community/embeddinggemma-2-bf16MLXfrom google/embeddinggemma-21.5 GB6,962
mlx-community/embeddinggemma-2-8bitMLXfrom google/embeddinggemma-21.3 GB6,514
mlx-community/Qwen3-Embedding-0.6B-8bitMLXfrom Qwen/Qwen3-Embedding-0.6B649 MB6,461
unum-cloud/uform3-image-text-multilingual-baseCore ML2 GB6,177
mlx-community/Qwen3-Embedding-4B-4bit-DWQMLXfrom Qwen/Qwen3-Embedding-4B—4,157
mlx-community/bge-m3-mlx-fp16MLXfrom BAAI/bge-m3—3,749
mlx-community/Qwen3-Embedding-8B-4bit-DWQMLXfrom Qwen/Qwen3-Embedding-8B—2,491
mlx-community/bge-m3-mlx-8bitMLXfrom BAAI/bge-m3—2,372
FluidInference/embeddinggemma-2-coremlCore MLfrom google/embeddinggemma-2—2,131
camstack/camstack-modelsCore ML—1,626
mlx-community/embeddinggemma-2-mxfp8MLXfrom google/embeddinggemma-2—1,279
mlx-community/embeddinggemma-2-4bitMLXfrom google/embeddinggemma-2—1,166
mlx-community/Qwen3-VL-Embedding-2B-8bitMLXfrom Qwen/Qwen3-VL-2B-Instruct—1,024
mlx-community/Qwen3-Embedding-8B-mxfp8MLXfrom Qwen/Qwen3-8B-Base—949
mlx-community/bge-small-en-v1.5-bf16MLX—832
mlx-community/Qwen3-Embedding-4B-mxfp8MLXfrom Qwen/Qwen3-4B-Base—801
mlx-community/CLM-v0.1-8B-MLX-8bitMLXfrom Qwen/Qwen3-8B—755
mlx-community/embeddinggemma-2-mxfp4MLXfrom google/embeddinggemma-2—742
mlx-community/Qwen3-Embedding-0.6B-mxfp8MLXfrom Qwen/Qwen3-0.6B-Base—697
unum-cloud/uform3-image-text-english-largeCore ML—676
jinaai/jina-embeddings-v5-text-small-retrieval-mlxMLXfrom jinaai/jina-embeddings-v5-text-small-retrieval—617
mlx-community/bge-m3-mlx-4bitMLXfrom BAAI/bge-m3—598
NateRunsA-LIST/alist-food-text-embedderCore MLfrom google/embeddinggemma-2—353
faxenoff/code-daemon-embed-v1Core MLfrom Alibaba-NLP/gte-modernbert-base—205
wolfram-pro/multilingual-e5-small-coremlCore MLfrom intfloat/multilingual-e5-small—157
DangerLabs/dm-jepa-1.5Core ML—155
yodayo-ai/jina-embeddings-v2-small-en-stCore ML—144
unum-cloud/uform3-image-text-english-smallCore ML—124
OpenInfinity/embeddinggemma-2-coreml-int8Core MLfrom google/embeddinggemma-2—107
FluidInference/granite-embedding-30m-sparse-coremlCore MLfrom ibm-granite/granite-embedding-30m-sparse—82
Arraasz/granite-embedding-small-english-r2-aneCore MLfrom ibm-granite/granite-embedding-small-english-r2—58
unum-cloud/uform3-image-text-english-baseCore ML—58
characteragent/granite-embedding-97m-multilingual-r2-coremlCore MLfrom ibm-granite/granite-embedding-97m-multilingual-r2—54
Severian/embedCore ML—50
haihengh/all-MiniLM-L6-v2-coremlCore ML—47
H4RUming/embeddinggemma-2-coreml-textCore MLfrom google/embeddinggemma-2—44
Nemosine/embeddinggemma-2-text-coremlCore MLfrom google/embeddinggemma-2—41
DangerLabs/dm-jepa-1.5-coremlCore ML—34
aufklarer/SpeechBrain-ECAPA-VoxCeleb-20M-CoreMLCore MLfrom speechbrain/spkrec-ecapa-voxceleb—32
apocryphx/embeddinggemma-300m-qat-q4_0-coremlCore MLfrom google/embeddinggemma-300m—29
JGKarlin/ohia-bge-m3-coremlCore MLfrom BAAI/bge-m3—23
Mannyking/embeddinggemma-2-text-270m-coremlCore MLfrom google/embeddinggemma-2—21
erjigit17/qwen3-embedding-0.6b-ane-coreml-8bitCore MLfrom neuradex/Qwen3-Embedding-0.6B-CoreML-ANE—20
usama99athar/multilingual-embedders-onnx-coremlCore ML—20
rankun203/granite-embedding-97m-multilingual-r2-coremlCore MLfrom ibm-granite/granite-embedding-97m-multilingual-r2—19
rankun203/granite-embedding-311m-multilingual-r2-coremlCore MLfrom ibm-granite/granite-embedding-311m-multilingual-r2—18

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