# NVIDIA API Catalog

> NVIDIA NIM — hosted, OpenAI-compatible model APIs for inference, vision, speech, retrieval, and more.

## Getting Started

All NVIDIA NIM APIs share a common base URL and auth scheme:

- **Base URL:** `https://integrate.api.nvidia.com/v1`
- **Auth:** Bearer token via `Authorization: Bearer $NVIDIA_API_KEY`
- **API Key:** Generate one at https://build.nvidia.com/settings

Example request:
```bash
curl https://integrate.api.nvidia.com/v1/chat/completions \
  -H "Authorization: Bearer $NVIDIA_API_KEY" \
  -d '{"model":"nvidia/llama-3.1-70b-instruct","messages":[{"role":"user","content":"Hello"}]}'
```

## Catalog

- [Models](/models.md) — NIM model APIs (LLMs, vision, speech, retrieval, and more)
- [Blueprints](/blueprints.md) — reference architectures and workflows
- [Skills](/skills.md) — prebuilt agent skills

## FAQ

**Which models are free?**
All models offer a free trial tier with no credit card required. Sign in to generate an API key.

**Which models are OpenAI API compatible?**
All NIM model endpoints at `integrate.api.nvidia.com/v1` implement the OpenAI Chat Completions API. Point an existing OpenAI SDK client at `https://integrate.api.nvidia.com/v1` and set the model name.

**How do I filter models by task type?**
Fetch /models.md — it lists all models with descriptions and supports `?page=N` for pagination.