Meta is introducing a new tool in the hopes of encouraging developers to utilize its range of Llama models for their upcoming projects. The company made the announcement during its first LlamaCon event in Menlo Park on Tuesday, unveiling the Llama API. This tool is now available as a limited free preview, enabling developers to experiment with Meta’s AI models, including the recently launched Llama 4 Scout and Maverick systems. Additionally, the Llama API simplifies the process of generating new API keys for authentication purposes.
In a blog post released during the event, the company stated, «We want to make it even easier for you to quickly start building with Llama, while also giving you complete control over your models and weights without being locked to an API.» The initial release of the Llama API includes tools that developers can utilize to refine and assess their applications.
Furthermore, Meta assured users that it will not utilize user prompts and model responses to train its own models. «When you’re ready, the models you create using the Llama API belong to you and can be hosted wherever you choose, as we do not retain them on our servers,» the company explained. Meta plans to expand the availability of the tool to more users in the upcoming weeks and months.
Despite Meta’s Llama models being downloaded over one billion times, the company is not typically regarded as a frontrunner in the AI sector in the same way as OpenAI and Anthropic. Meta’s reputation was further tarnished when it was discovered that the company manipulated LMArena to enhance the appearance of its Llama 4 models.
This article was originally published on Engadget at .
FUENTE