Latest SOTA models supported on Qualcomm NPU.
AI & ML interests
On Device AI Deployment and Research
Recent Activity
Text Generations Models in MLX format, hand picked by Nexa Team.
Text Generations Models in GGUF format, hand picked by Nexa Team.
Tiny, multimodal on-device models developed by Nexa AI.
Language Models that takes vision input and/or audio input, hand picked by Nexa Team.
-
NexaAI/gemma-3n-E4B-it-4bit-MLX
Image-Text-to-Text • Updated • 46 • 1 -
NexaAI/Qwen2.5-VL-7B-Instruct-4bit-MLX
Image-Text-to-Text • 2B • Updated • 36 -
NexaAI/SmolVLM-500M-Instruct-8bit-MLX
Image-Text-to-Text • 0.7B • Updated • 23 -
NexaAI/SmolVLM-Instruct-8bit-MLX
Image-Text-to-Text • 0.7B • Updated • 14
Language Models that takes vision input and/or audio input, hand picked by Nexa Team.
NexaQuant compresses models with 100% accuracy recovery.
Latest SOTA models supported on Qualcomm NPU.
Language Models that takes vision input and/or audio input, hand picked by Nexa Team.
-
NexaAI/gemma-3n-E4B-it-4bit-MLX
Image-Text-to-Text • Updated • 46 • 1 -
NexaAI/Qwen2.5-VL-7B-Instruct-4bit-MLX
Image-Text-to-Text • 2B • Updated • 36 -
NexaAI/SmolVLM-500M-Instruct-8bit-MLX
Image-Text-to-Text • 0.7B • Updated • 23 -
NexaAI/SmolVLM-Instruct-8bit-MLX
Image-Text-to-Text • 0.7B • Updated • 14
Text Generations Models in MLX format, hand picked by Nexa Team.
Language Models that takes vision input and/or audio input, hand picked by Nexa Team.
Text Generations Models in GGUF format, hand picked by Nexa Team.
NexaQuant compresses models with 100% accuracy recovery.
Tiny, multimodal on-device models developed by Nexa AI.