---
title: README
emoji: 🔗
colorFrom: gray
colorTo: gray
sdk: static
pinned: false
---

# LLM Tech

Inference API on GPUs in an Italian datacenter. Prompts and outputs are never written to disk.

Our quantized checkpoints, each measured against the original weights on the author's own evaluation set:

- [decider-4b-nvfp4](https://huggingface.co/llmtech/decider-4b-nvfp4)
- [decider-4b-fp8](https://huggingface.co/llmtech/decider-4b-fp8)
- [decider-2b-fp8](https://huggingface.co/llmtech/decider-2b-fp8)
- [decider-0.8b-fp8](https://huggingface.co/llmtech/decider-0.8b-fp8)

Models, pricing, API docs and a free trial key: https://llmtech.eu