netprojects/GLiDE-mlx-8bit
Image-Text-to-Text • 26B • Updated • 100 • 1
NetProjects MLX builds
Open models converted and tested for Apple Silicon (MLX) by NetProjects, each checked against the original model.
Note Fastino GLiDE 27B · calibrated decision model · 8-bit (recommended)
netprojects/GLiDE-mlx-4bit
Image-Text-to-Text • 26B • Updated • 102Note Fastino GLiDE 27B · calibrated decision model · 4-bit
netprojects/h2o-lightning-4b-mlx-q8
Image-Text-to-Text • 5B • Updated • 861 • 3Note H2O-Lightning 4B · small calibrated decision model, two-way reading · 8-bit (recommended)
netprojects/h2o-lightning-4b-mlx-q4
Image-Text-to-Text • 5B • Updated • 479 • 1Note H2O-Lightning 4B · small calibrated decision model · 4-bit, one-way
netprojects/granite-switch-4.1-8b-mlx-8bit
Text Generation • 8B • Updated • 168Note IBM Granite Switch 4.1.1 8B · chat, decision and guardian adapters · 8-bit
netprojects/granite-switch-4.1-8b-mlx-4bit
Text Generation • 9B • Updated • 142Note IBM Granite Switch 4.1.1 8B · 4-bit
netprojects/granite-switch-4.1-30b-mlx-8bit
Text Generation • 29B • Updated • 96Note IBM Granite Switch 4.1.1 30B · chat, decision and guardian adapters · 8-bit
netprojects/granite-switch-4.1-30b-mlx-4bit
Text Generation • 29B • Updated • 138Note IBM Granite Switch 4.1.1 30B · 4-bit
netprojects/AutoTrust-JEV-35B-mlx-8bit
Text Classification • 35B • Updated • 105Note AutoTrust JEV-35B · decision model with vision · 8-bit
netprojects/AutoTrust-JEV-35B-mlx-4bit-mixed
Text Classification • 35B • Updated • 116Note AutoTrust JEV-35B · 4-bit mixed
netprojects/AutoTrust-JEV-35B-mlx-3bit-mixed
Text Classification • 35B • Updated • 128Note AutoTrust JEV-35B · 3-bit mixed, the smallest
netprojects/AutoTrust-JEV-27B-v1.1-mlx-8bit
Text Classification • 27B • Updated • 130Note AutoTrust-JEV-27B v1.1 · decision model, bidirectional reading · 8-bit
netprojects/AutoTrust-JEV-27B-v1.1-mlx-4bit
Text Classification • 27B • Updated • 180Note AutoTrust-JEV-27B v1.1 · 4-bit
netprojects/nemotron-dory1.0-4b-mlx-8bit
Text Generation • 4B • Updated • 4.22k • 2Note NVIDIA Nemotron-Labs Dory 1.0 4B · looped reasoning LM · 8-bit
netprojects/nemotron-dory1.0-4b-mlx-bf16
Text Generation • 4B • Updated • 852Note NVIDIA Nemotron-Labs Dory 1.0 4B · bf16
netprojects/Nemotron-3.5-Embed-1B-mlx-bf16
Feature Extraction • 2B • Updated • 157Note NVIDIA Nemotron 3.5 Embed 1B: text + page-image embeddings, bf16
netprojects/Nemotron-3.5-Embed-1B-mlx-8bit
Feature Extraction • 2B • Updated • 66Note NVIDIA Nemotron 3.5 Embed 1B: text + page-image embeddings, 8-bit
netprojects/Nemotron-3.5-Embed-8B-mlx-8bit
Feature Extraction • 8B • Updated • 75Note NVIDIA Nemotron 3.5 Embed 8B: text + page-image embeddings, 8-bit
netprojects/Nemotron-3.5-Embed-8B-mlx-4bit
Feature Extraction • 8B • Updated • 86Note NVIDIA Nemotron 3.5 Embed 8B: text + page-image embeddings, 4-bit
netprojects/Nemotron-3.5-Rerank-8B-mlx-8bit
Text Classification • 8B • Updated • 64Note NVIDIA Nemotron 3.5 Rerank 8B: text + page-image reranker, 8-bit
netprojects/Nemotron-3.5-Rerank-8B-mlx-4bit
Text Classification • 8B • Updated • 77Note NVIDIA Nemotron 3.5 Rerank 8B: text + page-image reranker, 4-bit
netprojects/nemotron-dory1.0-4b-mlx-nvfp4
Text Generation • 4B • UpdatedNote 4-bit, from NVIDIA's NVFP4 checkpoint
netprojects/SenseNova-Loop-9B-Preview-MLX-8bit
Text Generation • 9B • UpdatedNote SenseNova-Loop-9B-Preview · looped LM with two-stream mHC · 8-bit (recommended)
netprojects/SenseNova-Loop-9B-Preview-MLX-6bit
Text Generation • 9B • UpdatedNote SenseNova-Loop-9B-Preview · 6-bit
netprojects/SenseNova-Loop-9B-Preview-MLX-4bit
Text Generation • 9B • UpdatedNote SenseNova-Loop-9B-Preview · 4-bit
netprojects/Cosmos3-Edge-mlx-8bit
Image-Text-to-Text • 2B • UpdatedNote NVIDIA Cosmos3-Edge reasoner · vision-language, text, image and video · 8-bit (recommended)
netprojects/Cosmos3-Edge-mlx-6bit
Image-Text-to-Text • 2B • UpdatedNote NVIDIA Cosmos3-Edge reasoner · 6-bit