byjg/llm-api-objects

Strongly-typed PHP objects for LLM API request payloads

6.0.0 2026-03-06 00:57 UTC

Requires

• php: >=8.3
• byjg/anydataset-xml: ^6.0

Requires (Dev)

• openai-php/client: ^v0.19
• phpunit/phpunit: ^10.5|^11.5
• vimeo/psalm: ^5.9|^6.13

MIT

README

Strongly-typed PHP model layer for building OpenAI Chat Completions API request payloads. Works with both OpenAI and Ollama (via the /v1 OpenAI-compatible endpoint) using the openai-php/client.

Features

• Strongly-typed models for Chat Completions requests
• Full support for function calling (tools): Tool, ToolFunction, ToolParameter, ToolChoice
• Multi-turn conversations with role=tool messages and tool_call_id
• Automatic message splitting when content exceeds a token limit
• Delimiter-based message partitioning
• Compatible with OpenAI and Ollama (/v1 endpoint)
• Ollama model tuning parameters via ModelFileParameters
• Jira XML/RSS importers to seed conversation context

Documentation

• Getting Started — Installation and first request
• Chat Model — Building and sending chat requests
• Messages — Roles, content, tool_call_id, tool_calls
• Tools — Function calling: Tool, ToolFunction, ToolParameter, ToolChoice
• Model Parameters — Ollama model tuning parameters
• Importers — Jira XML/RSS importers

Quick Examples

Basic chat with OpenAI

<?php
require 'vendor/autoload.php';

$client = OpenAI::factory()
    ->withApiKey($_ENV['OPENAI_API_KEY'])
    ->make();

$chat = new \ByJG\LlmApiObjects\Model\Chat(
    model: 'gpt-4o',
    messages: [
        new \ByJG\LlmApiObjects\Model\Message(
            role: \ByJG\LlmApiObjects\Enum\Role::user,
            message: 'What is the capital of France?',
        ),
    ],
    system: 'You are a helpful geography assistant.',
);

$result = $client->chat()->create($chat->toApi());
echo $result->choices[0]->message->content;

Basic chat with Ollama

<?php
$client = OpenAI::factory()
    ->withApiKey('ollama')
    ->withBaseUri('http://localhost:11434/v1')
    ->make();

$chat = new \ByJG\LlmApiObjects\Model\Chat(
    model: 'llama3',
    messages: [
        new \ByJG\LlmApiObjects\Model\Message(
            role: \ByJG\LlmApiObjects\Enum\Role::user,
            message: 'Hello!',
        ),
    ],
);

$result = $client->chat()->create($chat->toApi());
echo $result->choices[0]->message->content;

Function calling (tools)

<?php
use ByJG\LlmApiObjects\Enum\ToolChoice;
use ByJG\LlmApiObjects\Model\Chat;
use ByJG\LlmApiObjects\Model\Message;
use ByJG\LlmApiObjects\Enum\Role;
use ByJG\LlmApiObjects\Tool\Tool;
use ByJG\LlmApiObjects\Tool\ToolFunction;
use ByJG\LlmApiObjects\Tool\ToolParameter;

$weatherTool = new Tool(
    new ToolFunction(
        name: 'get_current_weather',
        description: 'Get the current weather in a given location',
        parameters: [
            new ToolParameter(name: 'location', type: 'string', description: 'City name', required: true),
            new ToolParameter(name: 'unit', type: 'string', enum: ['celsius', 'fahrenheit']),
        ],
    ),
);

$chat = new Chat(
    model: 'gpt-4o',
    messages: [
        new Message(role: Role::user, message: "What's the weather in Paris?"),
    ],
);
$chat->addTool($weatherTool);
$chat->setToolChoice(ToolChoice::auto);

$result = $client->chat()->create($chat->toApi());

Install

composer require byjg/llm-api-objects

Running Unit Tests

composer test

Dependencies

flowchart TD
    byjg/llm-api-objects --> byjg/anydataset-xml

Loading

Open source ByJG