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

# Llm ollama

## Overview

You can use Instructor with local Ollama instance.

Please note that, at least currently, OS models do not perform on par with OpenAI (GPT-3.5 or GPT-4) model for complex data schemas.

Supported modes:

* Instructor markdown-JSON fallback - fallback mode, works with any capable model
* native JSON object response\_format - recommended
* tool calling - supported (for selected models - check Ollama docs)

## Example

```php theme={null}
<?php

use Cognesy\Messages\Messages;
use Cognesy\Polyglot\Inference\Inference;
use Cognesy\Utils\Str;

require 'examples/boot.php';

$answer = Inference::using('ollama')
    ->with(
        messages: Messages::fromString('What is the capital of France'),
        options: ['max_tokens' => 64]
    )
    ->get();

echo "USER: What is capital of France\n";
echo "ASSISTANT: $answer\n";
assert(Str::contains($answer, 'Paris'));
?>
```