en.Wedoany.com Reported - Meta has released Muse Spark, the first AI model from its Super Intelligence Lab, which was established nearly a year ago. CEO Mark Zuckerberg stated that Meta will focus on developing personal superintelligence technology.

Muse Spark is the first product in the Muse model series, featuring multimodal reasoning capabilities, tool use support, visual chain-of-thought, and multi-agent orchestration. The model is now available via meta.ai and the Meta AI app, while Meta has also opened a private API preview to select users.

Muse Spark performs well in multimodal perception, reasoning, health, and agent tasks. However, Meta acknowledged performance gaps in coding and multi-step task-solving agent systems, and has invested funds to address these areas. Additionally, Meta introduced Contemplating mode, enabling Muse Spark to compete with reasoning modes from models like Gemini Deep Think and OpenAI's GPT Pro.

With Muse Spark, Meta has developed a model capable of understanding users' personal worlds, such as health information. Designed for personal AI use, the model primarily serves consumers and everyday users in health applications. At the same time, the social platform company seeks to become a leader in AI technology, and offering a personal-use model helps achieve this goal.

Lian Jye Su, an analyst at Omdia under Informa TechTarget, believes Meta has a clear strategy in product releases. The model is designed for daily tasks and interactions within Meta's application ecosystem, where it currently performs well. As Meta's first closed-source model (its open-source models are the Llama series), it faces competition from products by companies like Anthropic and OpenAI's Claude, potentially becoming another "me-too" product. However, Su noted that Meta's advantage lies in its vast data storage repository, and it will be the only AI product compatible with Meta's popular apps (WhatsApp, Facebook, and Instagram). The new Muse model series marks a fresh start for the company, which previously released open-source models. Targeting personal users may not be a disadvantage and could even extend Meta's influence into the enterprise sector.

Su added that if more people use Meta's platforms due to AI, all other businesses that advertise on or leverage these platforms to reach more users will benefit.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com