NexaAI Releases OmniNeural-4B for On-Device AI
The new 4-billion-parameter model is designed for 'any-to-any' multimodal tasks and optimized to run efficiently on mobile hardware.
NexaAI has introduced OmniNeural-4B, a new 4-billion-parameter model designed to bring versatile AI capabilities directly to consumer devices. Its compact size and specific optimizations target the growing demand for powerful, locally-run artificial intelligence.
Unlike models that handle specific input-output pairs like text-to-image, OmniNeural-4B is an 'any-to-any' multimodal system. This architecture allows it to process and generate a wide combination of data types, which could enable more complex and integrated on-device applications.
Designed for the Edge
The core focus of this release is performance on edge devices. NexaAI highlights that OmniNeural-4B is specifically optimized for the Neural Processing Units (NPUs) found in modern Snapdragon chips and Android hardware. This approach bypasses the need for cloud servers, potentially offering faster, more private, and offline-capable AI experiences directly on a user's phone.
The model is available on Hugging Face under a permissive Creative Commons BY 4.0 license, allowing developers and researchers to freely experiment with building sophisticated multimodal features into mobile applications.
Sources
- Visit
NexaAI/OmniNeural-4B
Hugging Face
More in Any-to-Any
Thinking Machines Debuts Inkling Small, a Compact Multimodal MoE
The Apache-2.0 model brings mixture-of-experts efficiency to image, audio, and text tasks in a smaller footprint.
KRAFTON releases A.X-K2 Raon speech MoE model
The game maker's new open model blends text-to-speech and speech recognition in a single 21B mixture-of-experts system with just 3B active parameters.
Microsoft's Mage-VL Streams Video Natively
A codec-native multimodal foundation model aims to understand live video and vision-language input in real time.
0 comments
No comments yet. Be the first to weigh in.