Weave Code
Ai Meta Platform Laravel Package
symfony/ai-meta-platform
Bridge for Meta's Llama models in Symfony AI, translating prompts into the correct Llama 3, 3.2, and 3.3 chat formats so you can use Meta's platform seamlessly within your Symfony AI applications.
Meta Platform is a bridge that connects Symfony AI to Meta Llama models, letting you use Llama's chat and completion capabilities through Symfony's unified AI interfaces. It handles the platform-specific details—such as model naming and the distinct prompt formats used across Llama generations—so you can focus on building AI-powered features rather than provider quirks.
- Seamless integration with the Symfony AI component ecosystem
- Support for Llama 3, 3.2, and 3.3 prompt formats
- Consistent, platform-agnostic API for chatting with Llama models
- Easy configuration and wiring within Symfony applications
Frequently asked questions about Ai Meta Platform
- How do I install symfony/ai-meta-platform in a Laravel application?
- Run composer require symfony/ai-meta-platform and make sure the Symfony AI platform contracts resolve in your Laravel project. The package is a library-level bridge, so Laravel will not auto-wire it for you. Add a ServiceProvider that binds the Meta platform and model clients, then read endpoint, model, and key values from config/services.php and .env. If you want a clean call site, add a thin facade or typed wrapper such as Llama::chat($messages).
- Is symfony/ai-meta-platform a Laravel package or a Symfony component?
- No, it is not a Laravel package. It is a Symfony AI platform bridge, one of the vendor bridges that implement Symfony AI's platform contracts alongside OpenAI, Anthropic, Gemini, and Ollama. Composer will install it into a Laravel app, but nothing registers services, publishes config, or provides a Laravel-native API. In Laravel it is a library dependency you integrate yourself, not a package with framework integration.
- Which Meta Llama models and prompt formats does this bridge support?
- The bridge supports Meta Llama 3, 3.2, and 3.3 prompt formats. Its main job is translating prompts into the correct Llama chat template, including the [INST] and [/INST] conventions and per-version differences, through Symfony AI's unified interfaces. That means you can keep application code platform-agnostic while the bridge handles Llama's model naming and formatting quirks.
- Do I need an API key or hosted Llama endpoint to use this package?
- Yes, practically speaking. Meta does not provide a single first-party hosted inference API with the same ergonomics as OpenAI, so you must supply an endpoint. That can be a local runtime such as Ollama or vLLM, or a hosted provider such as Amazon Bedrock, Together, Groq, or Fireworks. The bridge handles prompt formatting and contract conformance, not turnkey vendor hosting or credentials.
- How do I register and configure the Meta Llama platform in Laravel?
- Register the platform and model clients in a Laravel ServiceProvider, because Symfony's services.yaml autowiring conventions do not apply. Put base URL, model name, API key, temperature, and max tokens in config/services.php with matching .env entries. The bridge is unlikely to ship publishable Laravel config, so this configuration surface is yours to design. Keep the binding behind an interface so you can swap Llama endpoints or providers later.
- Can I use this bridge with Ollama, vLLM, Bedrock, Groq, Together, or Fireworks?
- Yes, provided the bridge can be pointed at that provider's Llama-compatible endpoint and the transport requirements match. Ollama and vLLM are common local options, while Bedrock, Together, Groq, and Fireworks offer hosted Llama inference. The bridge standardises Llama prompt formats, but you still need to verify base URL, authentication, model naming, and streaming support for each provider. Test one provider end-to-end before standardising on the bridge.
- Does symfony/ai-meta-platform support streaming, tool calling, structured output, or embeddings?
- The package's documented scope is chat prompt formatting for Llama 3, 3.2, and 3.3, not a full feature surface. Streaming, tool or function calling, structured JSON output, embeddings, token counting, retries, and timeouts are not clearly documented as part of this bridge. Some of that may come from Symfony AI's broader components, but you should verify each capability before committing. For a chat UI, confirm streaming works with SSE or Livewire before building around it.
- Does the package use PSR-18 or Symfony HttpClient, and can I swap in Laravel's HTTP client?
- The README does not document the HTTP transport, so do not assume you can substitute Laravel's HTTP client without reading the constructor signatures. If the bridge requires symfony/http-client, that package works under Laravel, or you can inject a Laravel Http-backed adapter if a PSR-18 client is accepted. Verify this early because transport choice affects testing, retries, and timeout handling. Treat substitutability as an integration task, not a given.
- Is symfony/ai-meta-platform production-ready and actively maintained?
- Be cautious. The package has effectively zero adoption signal: about 2 stars, 0 dependents, and no ecosystem references or evidence of production usage. Its release history shows eleven consecutive content-free minor bumps from 0.4 to 0.15, and v0.15.0 contains a single commit that only bumps the symfony/ai-platform constraint to 0.15. Under Symfony AI's 0.x policy that upstream minor may include breaking contract changes, so pin versions and test upgrades rather than assuming stability.
- What are the alternatives to symfony/ai-meta-platform for using Llama in Laravel?
- Alternatives depend on whether you want Symfony AI's PlatformInterface as your internal LLM abstraction. If you do, this bridge saves you writing two or three Llama chat templates and enables multi-model routing or swap-on-failure through the shared interface. If you do not, marginal value is close to zero, and direct provider SDKs, raw HTTP calls to Ollama or vLLM, or another Laravel AI package may be simpler. Evaluate the abstraction first, then choose the bridge.
Weaver
How can I help you explore Laravel packages today?