> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tokenlab.sh/llms.txt
> Use this file to discover all available pages before exploring further.

# Agno

> Use TokenLab models in Agno agents

## Overview

Agno can use TokenLab as an OpenAI-compatible model provider for agents and workflows.

<Note>
  **Type**: Agent Framework

  **Primary Path**: OpenAI-compatible chat completions

  **Support Confidence**: Supported path
</Note>

## Environment

```bash theme={null}
export TOKENLAB_API_KEY="sk-your-tokenlab-key"
```

## TokenLab Model Wrapper

If your Agno version includes the TokenLab model wrapper:

```python theme={null}
from agno.agent import Agent
from agno.models.tokenlab import TokenLab

agent = Agent(model=TokenLab(id="gpt-5.4-mini"))
agent.print_response("Give me a concise project risk checklist.")
```

Some Agno builds also support model strings:

```python theme={null}
agent = Agent(model="tokenlab:gpt-5.4-mini")
```

## OpenAI-Compatible Fallback

If your version does not include TokenLab yet, use Agno's OpenAI-compatible model path with:

* Base URL: `https://api.tokenlab.sh/v1`
* API key: `TOKENLAB_API_KEY`
* Model: `gpt-5.4-mini`

## Model Suggestions

Use `gpt-5.4-mini` for default agents, `gpt-5.4` or `claude-sonnet-5` for deeper reasoning, and `deepseek-v4-flash` for faster loops.