Skip to content
AI Model Radar
Apple silicon

Apple MLX models

5,841 models converted for Apple MLX, listed from Hugging Face’s mlx-community, refreshed 2h ago. Weights size and Mac memory are estimates from parameter count and quantisation.

Alert me about new Apple/MLX models

2,922 builds · page 12 of 49 · clear filters

ArrowNeo-AME-3x4B-v0.1-MoE-4bit

from DataPilot/ArrowNeo-AME-3x4B-v0.1-MoE
4-bit
Text generation · DataPilot ↓ 164 · ♥ 1
Params
7.8B
Weights
≈ 4.4 GB
Mac memory
8 GB+

mit · updated 2025-03-29

Qwen3-4B-Instruct-2507-5bit

from Qwen/Qwen3-4B-Instruct-2507
5-bit
Text generation · Qwen ↓ 164 · ♥ 1
Params
4.0B
Weights
≈ 2.8 GB
Mac memory
8 GB+

apache-2.0 · updated 2025-10-09

gemma-4-26B-A4B-it-qat-assistant-8bit

from google/gemma-4-26B-A4B-it-qat-q4_0-unquantized-assistant
8-bit
Text generation · Google · Gemma ↓ 164 · ♥ 0
Params
420M
Weights
≈ 446 MB
Mac memory
8 GB+

gemma · updated 2026-06-05

Llama-3_3-Nemotron-Super-49B-v1_5-mlx-4Bit

from unsloth/Llama-3_3-Nemotron-Super-49B-v1_5
4-bit
Text generation · unsloth ↓ 163 · ♥ 3
Params
50B
Weights
≈ 28.1 GB
Mac memory
48 GB+

nvidia-open-model-license · updated 2025-09-15

Llama-3.1-Nemotron-70B-Instruct-HF-4bit

from nvidia/Llama-3.1-Nemotron-70B-Instruct-HF
4-bit
Text generation · NVIDIA ↓ 163 · ♥ 2
Params
71B
Weights
≈ 39.7 GB
Mac memory
48 GB+

llama3.1 · updated 2024-10-17

ChatMusician-hf-4bit-mlx

mlx-community/ChatMusician-hf-4bit-mlx
4-bit
Text generation ↓ 163 · ♥ 0
Params
6.7B
Weights
≈ 4.0 GB
Mac memory
8 GB+

mit · updated 2024-03-01

kanana-2-3b-instruct-4bit

from kakaocorp/kanana-2-3b-instruct
4-bit
Text generation · kakaocorp ↓ 163 · ♥ 0
Params
3.2B
Weights
≈ 1.8 GB
Mac memory
8 GB+

kanana-open-license · updated 2026-08-06

mlx-mistral-7B-v0.1

mlx-community/mlx-mistral-7B-v0.1
Text generation ↓ 163 · ♥ 0
Params
7.2B
Weights
≈ 28.7 GB
Mac memory
48 GB+

apache-2.0 · updated 2024-02-08

Qwen2.5-32B-Instruct-8bit

from Qwen/Qwen2.5-32B
8-bit
Text generation · Qwen ↓ 162 · ♥ 7
Params
33B
Weights
≈ 34.8 GB
Mac memory
48 GB+

apache-2.0 · updated 2024-09-18

LLaDA2.2-flash-OptiQ-2bit

from inclusionAI/LLaDA2.2-flash
2-bit
Text generation · inclusionAI ↓ 162 · ♥ 4
Params
103B
Weights
≈ 32.2 GB
Mac memory
48 GB+

apache-2.0 · updated 2026-07-24

DeepSeek-V4-Flash-bf16

mlx-community/DeepSeek-V4-Flash-bf16
bf16
Text generation ↓ 162 · ♥ 2
Params
284B
Weights
≈ 1123 GB
Mac memory
> 512 GB

updated 2026-04-25

Josiefied-Qwen2.5-14B-Instruct-abliterated-v4-4-bit

from Goekdeniz-Guelmez/Josiefied-Qwen2.5-14B-Instruct-abliterated-v4
4-bit
Text generation · Goekdeniz-Guelmez ↓ 162 · ♥ 1
Params
15B
Weights
≈ 8.3 GB
Mac memory
16 GB+

apache-2.0 · updated 2024-12-11

Meta-Llama-3-8B-4bit

mlx-community/Meta-Llama-3-8B-4bit
4-bit
Text generation ↓ 161 · ♥ 8
Params
8.0B
Weights
≈ 5.3 GB
Mac memory
8 GB+

llama3 · updated 2024-04-20

MiniCPM5-1B-8bit

from openbmb/MiniCPM5-1B
8-bit
Text generation · openbmb ↓ 161 · ♥ 3
Params
1.1B
Weights
≈ 1.1 GB
Mac memory
8 GB+

apache-2.0 · updated 2026-05-26

Lamarck-14B-v0.7-4bit

from sometimesanotion/Lamarck-14B-v0.7
4-bit
Text generation · sometimesanotion ↓ 161 · ♥ 1
Params
15B
Weights
≈ 8.3 GB
Mac memory
16 GB+

apache-2.0 · updated 2025-02-19

Qwen2.5-Coder-3B-Instruct-bf16

from Qwen/Qwen2.5-Coder-3B-Instruct
bf16
Text generation · Qwen ↓ 161 · ♥ 1
Params
3.1B
Weights
≈ 6.2 GB
Mac memory
16 GB+

qwen-research · updated 2024-11-11

Strand-Rust-Coder-14B-v1-4bit

from Fortytwo-Network/Strand-Rust-Coder-14B-v1
4-bit
Text generation · Fortytwo-Network ↓ 161 · ♥ 1
Params
15B
Weights
≈ 8.3 GB
Mac memory
16 GB+

apache-2.0 · updated 2026-02-18

gemma-4-26B-A4B-it-qat-assistant-bf16

from google/gemma-4-26B-A4B-it-qat-q4_0-unquantized-assistant
bf16
Text generation · Google · Gemma ↓ 161 · ♥ 1
Params
420M
Weights
≈ 839 MB
Mac memory
8 GB+

gemma · updated 2026-06-05

phi-2

from microsoft/phi-2
Text generation · Microsoft · Phi ↓ 160 · ♥ 55
Params
—
Weights
—
Mac memory
—

microsoft-research-license · updated 2024-08-15

DeepSeek-R1-Distill-Qwen-32B-Japanese-4bit

from cyberagent/DeepSeek-R1-Distill-Qwen-32B-Japanese
4-bit
Text generation · cyberagent ↓ 160 · ♥ 2
Params
33B
Weights
≈ 18.4 GB
Mac memory
24 GB+

mit · updated 2025-01-27

Qwen3-4B-Sky-High-Hermes-8bit

from ZeroXClem/Qwen3-4B-Sky-High-Hermes
8-bit
Text generation · ZeroXClem ↓ 160 · ♥ 2
Params
4.0B
Weights
≈ 4.3 GB
Mac memory
8 GB+

apache-2.0 · updated 2026-01-19

Llama-3.1-Nemotron-70B-Instruct-HF-8bit

from nvidia/Llama-3.1-Nemotron-70B-Instruct-HF
8-bit
Text generation · NVIDIA ↓ 160 · ♥ 1
Params
71B
Weights
≈ 75.0 GB
Mac memory
96 GB+

llama3.1 · updated 2024-10-17

Meta-Llama-3.1-405B-Instruct-8bit

from meta-llama/Llama-3.1-405B-Instruct
8-bit
Text generation · Meta · Llama ↓ 160 · ♥ 1
Params
406B
Weights
≈ 431 GB
Mac memory
512 GB+

llama3.1 · updated 2024-11-10

japanese-stablelm-base-gamma-7b-4bit

mlx-community/japanese-stablelm-base-gamma-7b-4bit
4-bit
Text generation ↓ 160 · ♥ 0
Params
7.2B
Weights
≈ 4.1 GB
Mac memory
8 GB+

apache-2.0 · updated 2024-08-30

Phi-3-mini-4k-instruct-unsloth-4bit

mlx-community/Phi-3-mini-4k-instruct-unsloth-4bit
4-bit
Text generation ↓ 159 · ♥ 2
Params
3.8B
Weights
≈ 2.1 GB
Mac memory
8 GB+

mit · updated 2024-05-24

Ministral-3-3B-Instruct-2512

from mistralai/Ministral-3-3B-Instruct-2512
Text generation · Mistral ↓ 159 · ♥ 1
Params
3.4B
Weights
≈ 6.9 GB
Mac memory
16 GB+

apache-2.0 · updated 2025-12-03

GLM-Z1-9B-0414-4bit

from zai-org/GLM-Z1-9B-0414
4-bit
Text generation · zai-org ↓ 158 · ♥ 3
Params
9.4B
Weights
≈ 5.3 GB
Mac memory
8 GB+

mit · updated 2025-04-17

Josiefied-Qwen2.5-14B-Instruct-abliterated-v4-8-bit

from Goekdeniz-Guelmez/Josiefied-Qwen2.5-14B-Instruct-abliterated-v4
8-bit
Text generation · Goekdeniz-Guelmez ↓ 158 · ♥ 1
Params
15B
Weights
≈ 15.7 GB
Mac memory
24 GB+

apache-2.0 · updated 2024-12-09

MN-12B-Mag-Mell-R1-6bit

from inflatebot/MN-12B-Mag-Mell-R1
6-bit
Text generation · inflatebot ↓ 158 · ♥ 1
Params
12B
Weights
≈ 10.0 GB
Mac memory
16 GB+

updated 2025-05-18

BTL-4-OptiQ-4bit-REAP-18B

from mlx-community/BTL-4-OptiQ-4bit
4-bit
Text generation ↓ 158 · ♥ 0
Params
19B
Weights
≈ 10.4 GB
Mac memory
16 GB+

apache-2.0 · updated 2026-09-15

Llama-3.1-70B-Japanese-Instruct-2407-4bit

mlx-community/Llama-3.1-70B-Japanese-Instruct-2407-4bit
4-bit
Text generation ↓ 158 · ♥ 0
Params
71B
Weights
≈ 39.7 GB
Mac memory
48 GB+

llama3.1 · updated 2024-07-26

Llama-3.1-Nemotron-Nano-4B-v1.1-8bit

from nvidia/Llama-3.1-Nemotron-Nano-4B-v1.1
8-bit
Text generation · NVIDIA ↓ 158 · ♥ 0
Params
4.5B
Weights
≈ 4.8 GB
Mac memory
8 GB+

nvidia-open-model-license · updated 2025-06-04

Unsloth-DeepSeek-R1-Distill-Qwen-14B-4bit

from unsloth/DeepSeek-R1-Distill-Qwen-14B
4-bit
Text generation · unsloth ↓ 157 · ♥ 1
Params
15B
Weights
≈ 8.3 GB
Mac memory
16 GB+

apache-2.0 · updated 2025-02-26

gemma-2-9b-it-8bit

mlx-community/gemma-2-9b-it-8bit
8-bit
Text generation ↓ 156 · ♥ 3
Params
9.2B
Weights
≈ 9.8 GB
Mac memory
16 GB+

gemma · updated 2024-11-06

MiMo-7B-RL-4bit

from XiaomiMiMo/MiMo-7B-RL
4-bit
Text generation · XiaomiMiMo ↓ 156 · ♥ 1
Params
7.6B
Weights
≈ 4.3 GB
Mac memory
8 GB+

mit · updated 2025-05-02

LFM2.5-24B-A2B-Preview-mxfp8

from lq-preview/LFM2.5-24B-A2B-Preview
MXFP8
Text generation · lq-preview ↓ 155 · ♥ 0
Params
24B
Weights
≈ 25.5 GB
Mac memory
32 GB+

lfm1.0 · updated 2026-10-13

MiniMax-M2.7-4bit-mxfp4

from MiniMaxAI/MiniMax-M2.7
MXFP4
Text generation · MiniMaxAI ↓ 154 · ♥ 10
Params
229B
Weights
≈ 129 GB
Mac memory
192 GB+

modified-mit · updated 2026-04-12

Meta-Llama-3-70B-4bit

mlx-community/Meta-Llama-3-70B-4bit
4-bit
Text generation ↓ 154 · ♥ 9
Params
71B
Weights
≈ 39.7 GB
Mac memory
48 GB+

llama3 · updated 2024-04-20

Kimi-Dev-72B-5bit

from moonshotai/Kimi-Dev-72B
5-bit
Text generation · Moonshot · Kimi ↓ 154 · ♥ 2
Params
73B
Weights
≈ 50.0 GB
Mac memory
64 GB+

mit · updated 2025-07-11

MiniMax-M2.1-8bit

from MiniMaxAI/MiniMax-M2.1
8-bit
Text generation · MiniMaxAI ↓ 154 · ♥ 2
Params
229B
Weights
≈ 243 GB
Mac memory
512 GB+

modified-mit · updated 2026-02-18

medgemma-27b-text-it-8bit

from google/medgemma-27b-text-it
8-bit
Text generation · Google · Gemma ↓ 154 · ♥ 2
Params
28B
Weights
≈ 30.2 GB
Mac memory
48 GB+

health-ai-developer-foundations · updated 2025-06-09

Nemotron-Cascade-2-30B-A3B-nvfp4

from nvidia/Nemotron-Cascade-2-30B-A3B
4-bit
Text generation · NVIDIA ↓ 153 · ♥ 1
Params
32B
Weights
≈ 17.8 GB
Mac memory
24 GB+

nvidia-open-model-license · updated 2026-04-30

GLM-4-32B-0414-4bit

from THUDM/GLM-4-32B-0414
4-bit
Text generation · THUDM ↓ 152 · ♥ 4
Params
33B
Weights
≈ 18.3 GB
Mac memory
24 GB+

mit · updated 2025-04-21

Apertus-8B-Instruct-2509-4bit

from swiss-ai/Apertus-8B-Instruct-2509
4-bit
Text generation · swiss-ai ↓ 152 · ♥ 1
Params
8.1B
Weights
≈ 4.5 GB
Mac memory
8 GB+

apache-2.0 · updated 2025-09-03

Qwen3-4B-4bit-DWQ-053125

from Qwen/Qwen3-4B
4-bit DWQ
Text generation · Qwen ↓ 151 · ♥ 2
Params
4.0B
Weights
≈ 2.3 GB
Mac memory
8 GB+

apache-2.0 · updated 2025-06-01

DeepSeek-R1-Distill-Qwen-32B-Japanese

from cyberagent/DeepSeek-R1-Distill-Qwen-32B-Japanese
Text generation · cyberagent ↓ 151 · ♥ 1
Params
33B
Weights
≈ 65.5 GB
Mac memory
96 GB+

mit · updated 2025-01-27

starcoder2-3b-4bit

mlx-community/starcoder2-3b-4bit
4-bit
Text generation ↓ 151 · ♥ 1
Params
3.0B
Weights
≈ 1.9 GB
Mac memory
8 GB+

bigcode-openrail-m · updated 2024-03-01

NVIDIA-Nemotron-3-Super-120B-A12B-8bit

from nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16
8-bit
Text generation · NVIDIA ↓ 151 · ♥ 0
Params
121B
Weights
≈ 128 GB
Mac memory
192 GB+

nvidia-nemotron-open-model-license · updated 2026-03-31

Qwen2.5-Coder-3B-8bit

from Qwen/Qwen2.5-Coder-3B
8-bit
Text generation · Qwen ↓ 151 · ♥ 0
Params
3.1B
Weights
≈ 3.3 GB
Mac memory
8 GB+

qwen-research · updated 2024-11-11

NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-MXFP4

from nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16
MXFP4
Text generation · NVIDIA ↓ 150 · ♥ 3
Params
4.9B
Weights
≈ 2.8 GB
Mac memory
8 GB+

nvidia-open-model-license · updated 2025-12-17

Kimi-Dev-72B-8bit

from moonshotai/Kimi-Dev-72B
8-bit
Text generation · Moonshot · Kimi ↓ 150 · ♥ 2
Params
73B
Weights
≈ 77.3 GB
Mac memory
96 GB+

mit · updated 2025-07-12

Qwen2.5-14B-Instruct-8bit

from Qwen/Qwen2.5-14B
8-bit
Text generation · Qwen ↓ 150 · ♥ 2
Params
15B
Weights
≈ 15.7 GB
Mac memory
24 GB+

apache-2.0 · updated 2024-09-18

Hermes-3-Llama-3.2-3B-8bit

from NousResearch/Hermes-3-Llama-3.2-3B
8-bit
Text generation · NousResearch ↓ 150 · ♥ 1
Params
3.2B
Weights
≈ 3.4 GB
Mac memory
8 GB+

llama3 · updated 2024-12-11

Qwen2.5-3B-Instruct-8bit

from Qwen/Qwen2.5-3B
8-bit
Text generation · Qwen ↓ 150 · ♥ 0
Params
3.1B
Weights
≈ 3.3 GB
Mac memory
8 GB+

qwen-research · updated 2024-09-18

Laguna-S-2.1-oQ2e

from poolside/Laguna-S-2.1
2-bit
Text generation · poolside ↓ 149 · ♥ 12
Params
118B
Weights
≈ 36.8 GB
Mac memory
48 GB+

openmdw-1.1 · updated 2026-07-24

Hy3-OptiQ-2bit

from tencent/Hy3
2-bit
Text generation · tencent ↓ 149 · ♥ 6
Params
295B
Weights
≈ 92.3 GB
Mac memory
128 GB+

apache-2.0 · updated 2026-09-15

gemma-4-12b-coder-fable5-composer2.5

from yuxinlu1/gemma-4-12B-coder-fable5-composer2.5-v1
Text generation · yuxinlu1 ↓ 149 · ♥ 5
Params
12B
Weights
≈ 23.8 GB
Mac memory
32 GB+

apache-2.0 · updated 2026-06-21

Qwen2.5-7B-Instruct-1M-abliterated-4bit

from huihui-ai/Qwen2.5-7B-Instruct-1M-abliterated
4-bit
Text generation · huihui-ai ↓ 149 · ♥ 0
Params
7.6B
Weights
≈ 4.3 GB
Mac memory
8 GB+

apache-2.0 · updated 2025-02-19

Qwen3-235B-A22B-Instruct-2507-5bit

from Qwen/Qwen3-235B-A22B-Instruct-2507
5-bit
Text generation · Qwen ↓ 148 · ♥ 1
Params
235B
Weights
≈ 162 GB
Mac memory
192 GB+

apache-2.0 · updated 2025-07-22