Skip to main content
What you’ll build: A minimal Agno agent that talks to a SambaCloud-hosted model in under a dozen lines of Python.

​
Overview

Agno is a lightweight Python framework for building multi-modal AI agents. Wire it to SambaCloud and you get fast, low-ceremony agents that can:

Reason and respond

Stream natural-language replies from any SambaCloud-hosted LLM through a clean Agent interface.

Use tools

Compose decorator-based tools with full Python type hints — Agno generates the schemas for you.

Stay model-agnostic

Swap models with a single id= argument — same code, different backend.

Ship fast

Run as a script, a notebook cell, or behind a FastAPI service with minimal framework overhead.

​
Prerequisites

  • A SambaCloud account and API key from the SambaCloud portal
  • Python 3.9 or later installed. Confirm with python --version.
  • Familiarity with Python virtual environments and terminal basics

​
Setup

1

Create and activate a virtual environment

2

Install Agno and the OpenAI SDK

Agno’s Sambanova model is built on OpenAILike and uses the openai package under the hood. pip install agno alone is not enough — install both.
3

Export your SambaCloud API key

Never commit your API key to source control. Use a .env file or your shell’s secret manager.

​
Example

Use Sambanova as your Agent’s model. Always pass id= explicitly — the library’s default model is deprecated on SambaCloud.
horror_agent.py
Run it:
Swap id= for any model on the SambaCloud supported models page — the rest of the code stays the same.

​
Parameters

The Sambanova model class accepts the following arguments:

​
Additional resources

Agno repository

Source code, examples, and release notes for the Agno framework.

Agno documentation

Official guides covering agents, tools, multi-modal inputs, and storage.

Supported models

Current list of models available on SambaCloud.

Deprecations

Models scheduled for removal and their recommended replacements.

​
Troubleshooting

  • Install both packages: pip install agno openai
  • Agno’s Sambanova model class depends on openai under the hood — pip install agno alone is not enough
  • Confirm your virtual environment is active: which python should point inside .venv
  • Confirm the key is exported: echo $SAMBANOVA_API_KEY
  • Re-export with export SAMBANOVA_API_KEY="your-key" and retry
  • The Sambanova class reads SAMBANOVA_API_KEY automatically — no need to pass api_key= explicitly unless overriding
  • The library default id (Meta-Llama-3.1-8B-Instruct) is deprecated on SambaCloud — always pass id= explicitly
  • Check the SambaCloud models page for current model IDs
  • See Deprecations for recommended replacements
  • Model names are case-sensitive — use the exact string from the SambaCloud models page
  • Verify your SambaCloud account has access to the requested model