# OYD: On Your Device Super Intelligence

> OYD (oyd.si) is the local AI lab of WebEdge, Lithuania. We choose hardware, deploy and tune open language models on a company's own infrastructure, and publish our own benchmarks.

## Pages
- [Home](https://oyd.si/): services, benchmark highlights, process, FAQ, contact
- [Benchmarks](https://oyd.si/benchmarks/): measured local LLM speed (tok/s) with exact hardware, model and settings, plus the platforms we test on
- [Privacy policy](https://oyd.si/privacy-policy/)
- Lithuanian: https://oyd.si/lt/ · Russian: https://oyd.si/ru/

## Services
- Hardware selection and benchmarks (GPU servers, NVIDIA DGX Spark, Apple Silicon)
- Deployment and configuration (exllamav3, vLLM, llama.cpp, MLX; OpenAI-compatible API)
- Model selection and optimization (EXL3/GGUF quantization, speculative decoding, long context)
- Wiring models into work (document search, chat and coding assistants, API integrations)

## Contact
- Email: info@webedge.dev
- Parent company: WebEdge, https://webedge.dev/