On April 29, 2025, Meta launched the Llama API during its inaugural LlamaCon AI developer conference in Menlo Park.

This new API allows developers to create tools and applications that leverage Llama models, enhancing their development capabilities.

The Llama API is designed to work seamlessly with Meta's Software Development Kits (SDKs), simplifying the integration process.

It includes features for generating new API keys, fine-tuning models, and evaluating applications, making it a comprehensive tool for developers.

The API is expected to accelerate AI development for businesses and developers, helping them create more efficient applications.

This strategic move positions Meta to compete effectively in the open model market against rivals like DeepSeek and Alibaba's Qwen.

By launching the Llama API, Meta aims to strengthen its competitive edge against major players such as OpenAI and Google.

Despite the popularity of Llama models, which have been downloaded over a billion times, Meta's reputation in the AI sector has been marred by past controversies.

These controversies include allegations of manipulating benchmark tests to enhance the perceived performance of its Llama 4 models.

Meta plans to gradually expand access to the Llama API in the coming weeks and months, inviting more developers to participate.

Customers using the API can generate data, train their models, and utilize an evaluation suite to assess the quality of their customized models.

However, pricing details for the Llama API have not yet been disclosed, leaving potential users awaiting further information.