README.md

---
title: README
emoji: 📄
colorFrom: blue
colorTo: indigo
sdk: static
pinned: false
---

# Providus AI

We build document AI that runs where your documents live: on your own servers or in our cloud.
Our open models read document pages (text, headings, tables and charts) and turn them into clean Markdown.

## providus-vl-mini

An open-weight document parsing model, a fine-tune of MinerU2.5-Pro by OpenDataLab (Apache-2.0).

- **78.96 ± 0.04 on ParseBench** (mean of 4 full runs). On ParseBench's leaderboard (b9ba246, 2026-10-12) it would place first among open-weight models.
- **Zero per-page fees:** providus-vl-mini is free to use under Apache-2.0. Your only cost is your own hardware.

| Repo | What it is |
|---|---|
| [providus-vl-mini](https://huggingface.co/ProvidusAI/providus-vl-mini) | The full model, for vLLM on a GPU server (access approved automatically) |
| [providus-vl-mini-GGUF](https://huggingface.co/ProvidusAI/providus-vl-mini-GGUF) | One file for LM Studio, llama.cpp and Ollama (open, no login) |
| [providus-parse-tools](https://github.com/ProvidusAI/providus-parse-tools) | The pipeline that turns pages into Markdown (`pip install providus-parse-tools`) |

## Try it on your laptop (LM Studio)

```
lms get https://huggingface.co/ProvidusAI/providus-vl-mini-GGUF
lms load providus-vl-mini --context-length 8192 --identifier providus-vl-mini
lms server start
pip install providus-parse-tools==1.0.0
```

Then point the pipeline at `http://127.0.0.1:1234`. The model card has the full steps for LM Studio, llama.cpp and Ollama.

If you like our work, please follow us here, like the models and star the repo on GitHub.

[providus.ai](https://providus.ai) · [GitHub](https://github.com/ProvidusAI)