Strands Nvidia NIM Provider

A minimal custom provider that fixes message formatting compatibility between Strands Agents SDK and Nvidia NIM API.

Problem Solved

The standard Strands LiteLLM integration fails with Nvidia NIM because:

• Strands formats messages as structured content: [{"text": "content", "type": "text"}]
• Nvidia NIM expects simple string content: "content"

This provider bridges that gap by converting structured content to simple strings.

Installation

Option 1: Install from PyPI (Recommended)

pip install strands-nvidia-nim

Option 2: Install from GitHub

pip install git+https://github.com/yourusername/strands-nvidia-nim.git

Option 3: Local Development

git clone https://github.com/yourusername/strands-nvidia-nim.git
cd strands-nvidia-nim
pip install -e .

Quick Start

Option 1: Direct API Key

from strands import Agent
from strands_tools import calculator
from strands_nvidia_nim import NvidiaNIM

# Create the provider
model = NvidiaNIM(
    api_key="your-nvidia-nim-api-key",
    model_id="meta/llama-3.1-70b-instruct",
    params={
        "max_tokens": 1000,
        "temperature": 0.7,
    }
)

# Use with standard Strands Agent
agent = Agent(model=model, tools=[calculator])
agent("What is 123.456 * 789.012?")

Option 2: Environment Variables (Recommended)

# Set your API key as an environment variable
export NVIDIA_NIM_API_KEY=your-nvidia-nim-api-key

import os
from strands import Agent
from strands_tools import calculator
from strands_nvidia_nim import NvidiaNIM

model = NvidiaNIM(
    api_key=os.getenv("NVIDIA_NIM_API_KEY"),
    model_id="meta/llama-3.1-70b-instruct",
    params={"max_tokens": 1000, "temperature": 0.7}
)

agent = Agent(model=model, tools=[calculator])
agent("What is 123.456 * 789.012?")

Available Models

Popular Nvidia NIM models:

• meta/llama-3.1-70b-instruct - High quality, larger model
• meta/llama-3.1-8b-instruct - Faster, smaller model
• meta/llama-3.3-70b-instruct - Latest Llama model
• mistralai/mistral-large - Mistral's flagship model
• nvidia/llama-3.1-nemotron-70b-instruct - Nvidia-optimized

Configuration

model = NvidiaNIM(
    api_key="your-api-key",
    model_id="meta/llama-3.1-70b-instruct",
    params={
        "max_tokens": 1500,
        "temperature": 0.7,
        "top_p": 0.9,
        "frequency_penalty": 0.0,
        "presence_penalty": 0.0
    }
)

Features

• ✅ Fixes BadRequestError - No more message formatting issues
• ✅ Minimal Code - Simple, focused solution
• ✅ Standard Strands Pattern - Uses Agent(model=model, tools=[tools])
• ✅ Clean Output - Proper streaming without artifacts
• ✅ Error Handling - Context window overflow detection

Comparison

Contributing

We welcome contributions! Please see CONTRIBUTING.md for guidelines.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Support

• 🐛 Bug Reports: GitHub Issues
• 💬 Discussions: GitHub Discussions
• 📖 Documentation: README