QVAC

QVAC

5|0 reviews|0 favorites
Visit website
Introduction
Tether's open-source, decentralized local AI SDK that runs text, vision and audio models entirely on-device with no cloud or API keys.
Listed on
September 12, 2026
FreeAPIAI DatasetsAI FrameworksAI SDKAI Computer Vision ProjectsAI Mobile DevelopmentAI Model DeploymentAI Model Fine-tuningAI Open Source LLMAI RAG PlatformSpeech AI Projects
QVAC product preview
Visit website

What is QVAC?

QVAC is an open-source local AI ecosystem from Tether that lets intelligence run privately and offline on any device, from phones and laptops to robots and wearables. Its core is the QVAC SDK, a single JavaScript or Python API covering 10+ AI tasks such as text generation, embeddings, RAG, transcription, translation, image generation, TTS, OCR, vision and fine-tuning. The ecosystem also includes QVAC Fabric for on-device LoRA fine-tuning, QVAC Psy specialized open-weight models, and QVAC Genesis, an open synthetic STEM reasoning dataset. Everything is Apache 2.0 licensed and free, with no per-token billing or API keys.

How to use QVAC?

  • Install the QVAC SDK in JavaScript or Python with a single install, then load a model (local or from Hugging Face) and call tasks like completion, embeddings, RAG, transcription or image generation through one unified interface. For customization, use QVAC Fabric to train a LoRA adapter on your own data directly on-device, which returns a portable .gguf adapter you can ship and load anywhere. Models can be run on iOS, Android, desktop or server hardware, and peer-to-peer offload is supported. Start via the Get started page or the GitHub repository.

Core features of QVAC

  • Single unified API for 10+ on-device AI tasks (text, vision, audio)
  • Fully offline inference with no API keys, per-token cost or rate limits
  • Privacy by architecture: prompts and files never leave the device
  • On-device LoRA fine-tuning via QVAC Fabric, even on mobile GPUs
  • Vulkan and Metal GPU backends, no CUDA requirement
  • Cross-platform: iOS, Android, desktop and server with one API
  • Peer-to-peer model offload across devices
  • Open-weight specialized models (QVAC Psy) and open datasets (QVAC Genesis)
  • Apache 2.0 open source

User reviews

5.0

/ 5

0 reviews

5 stars
0
4 stars
0
3 stars
0
2 stars
0
1 stars
0

No reviews yet. Be the first to write one.

Similar products

QVAC Embed

Add a website badge to show your product on Hootool—help others find the right AI tool for their problem. Easy to place on your homepage or footer.

Featured on

Hootool.ai