---
title: NVIDIA NIM
description: Use NVIDIA NIM for GPU-optimized inference and embeddings with Nexus.
---

NVIDIA NIM provides GPU-optimized inference with embedding support. Uses an OpenAI-compatible API.

## Installation

```go
import "github.com/xraph/nexus/providers/nvidia"
```

## Quick Start

```go
provider := nvidia.New(os.Getenv("NVIDIA_API_KEY"))

gw := nexus.New(
    nexus.WithProvider(provider),
)
```

## Options

| Option | Description |
|--------|-------------|
| `nvidia.WithBaseURL(url)` | Override the API base URL (default: `https://integrate.api.nvidia.com/v1`) |

## Capabilities

| Capability | Supported |
|-----------|-----------|
| Chat | Yes |
| Streaming | Yes |
| Embeddings | Yes |
| Vision | No |
| Tools | Yes |
| Thinking | No |

## Models

| Model | Context | Max Output | Input Price | Output Price |
|-------|---------|-----------|-------------|-------------|
| `meta/llama-3.1-405b-instruct` | 131K | 4,096 | $5.00/M | $16.00/M |
| `meta/llama-3.1-8b-instruct` | 131K | 4,096 | $0.30/M | $0.50/M |
| `nvidia/nv-embedqa-e5-v5` | 512 | — | $0.02/M | — |