Weave Code
Ai Ollama Platform Laravel Package
symfony/ai-ollama-platform
Symfony AI bridge for the Ollama platform. Connect Symfony AI to Ollama’s chat and embedding APIs, including NDJSON streaming, using Ollama models and Modelfile capabilities. Links to docs, issues, and contributions in the main Symfony AI repo.
Ollama platform bridge for Symfony AI
Frequently asked questions about Ai Ollama Platform
- Can I use symfony/ai-ollama-platform directly in Laravel without Symfony dependencies?
- No, this package is Symfony-first and requires abstraction. You’ll need to replace Symfony’s HttpClient with Laravel’s Http facade or Guzzle, and adapt Messenger for Laravel Queues. The core Ollama API abstraction (chat/embeddings/streaming) remains usable with minimal effort.
- How do I handle NDJSON streaming in Laravel for real-time chatbots?
- Laravel’s synchronous lifecycle may need adjustments. Use Swoole, ReactPHP, or Laravel Queues to process streaming responses asynchronously. For WebSocket-based apps, pair with Laravel Echo/Pusher to broadcast events in real-time.
- What Laravel versions support symfony/ai-ollama-platform?
- The package targets PHP 8.1+, so Laravel 9+ is recommended. For older versions, manually upgrade dependencies or use Docker with `php:8.1`. Test compatibility with Laravel Pint and PHPStan to catch version-specific issues.
- Is there a simpler Laravel-native alternative to this Symfony package?
- Yes, consider `olliwood/ollama-php`, a lightweight Laravel-specific client for Ollama’s HTTP API. It avoids Symfony dependencies entirely but lacks advanced features like Modelfile support or structured DeltaInterface outputs.
- How do I manage Ollama models (updates, scaling) in a Laravel production environment?
- Use Docker/Kubernetes for Ollama infrastructure and implement Laravel Artisan commands to sync models. Cache model metadata in Laravel’s database or Redis. For scaling, leverage Ollama’s native capabilities or integrate with a model registry service.
- Can I use this package for multimodal AI (e.g., audio models like gemma:2b) in Laravel?
- Yes, but test thoroughly in staging first. The package supports Ollama’s audio capabilities, but Laravel may need custom event listeners or middleware to handle binary data streams. Validate outputs with Laravel’s validation rules or pipes.
- How do I replace Symfony’s Messenger component for async tasks in Laravel?
- Replace Messenger with Laravel Queues or Horizon. Wrap the package’s async methods in a Laravel Queue job, using `dispatch()` or `dispatchSync()`. For event-driven workflows, emit Laravel events and listen with `Bus::dispatch()`.
- Will this package work with Laravel’s service container and dependency injection?
- Yes, but you’ll need to bind Symfony services to Laravel interfaces. For example, create a `OllamaClientInterface` and bind it to a Laravel service provider. The Provider abstraction in the package helps isolate dependencies.
- How do I handle errors or retries for Ollama API calls in Laravel?
- Use Laravel’s HTTP client retry logic or middleware. For example, wrap the Ollama client in a custom middleware that retries failed requests with exponential backoff. Log errors via Laravel’s logging system for observability.
- Can I swap Ollama for another LLM provider (e.g., local models, cloud APIs) later?
- The Provider abstraction allows backend swapping, but you’ll need Laravel-specific adapters. For example, create a `LocalLLMProvider` interface and implement it for non-Ollama services. Cache model catalogs in Laravel’s database to reduce API calls.
Weaver
How can I help you explore Laravel packages today?