Meta company (LLAMA-5) has launched its first artificial intelligence model named "Muse Spark". This model was developed by a star team called "MSL", whose engineers and researchers were attracted by billion-dollar offers.
The editorial office reports that the model's comparative test results and main functionalities are reflected in various indicators. These include a multi-agent mode called "Contemplating", the effectiveness of initial training, and "HLE" quality depending on the number of agents and user latency. The "Contemplating" mode offers functionality similar to the operating principles of models like "Deep Think" or "GPT Pro".
No information has yet been provided regarding the full release of the model and the public dissemination of its weights. However, "LLAMA-5 Muse Spark" will be accessible to selected companies via a closed API and is already available in the company's artificial intelligence services. Users can interact with the model at Meta AI.
Meta's CEO of Artificial Intelligence, Aleksandr Wang, has made a statement about future plans. He announced that larger models have already been trained and will be launched soon.