Meta launches Muse Spark AI model to replace Llama across WhatsApp, Instagram, and Facebook

Meta's Muse Spark: a smaller, faster AI model for broad app deployment

Meta's new "small and fast" AI model, Muse Spark, is an acknowledgement that as enterprises scale AI systems beyond millions of users and for use on a greater variety of devices, they must make things more efficient and more application-specific. Muse Spark now powers the Meta AI assistant on the web and in the Meta AI app, and the company plans to roll it out across WhatsApp, Instagram, Facebook, Messenger, and the company's smart glasses. It will also offer select partners access to the underlying technology through an API, initially as a private preview. "We hope to open-source future versions of the model," it said in a blog post announcing Muse Spark. While Meta did not disclose the model's size or much about its architecture, it described Muse Spark as being capable of balancing capability with speed.

So long, Llama: Meta unveils Muse Spark AI with Contemplating mode and health focus - 9to5Mac

Llama is out; Muse Spark is in. Meta has unveiled its rebooted AI efforts with a new model it calls Muse Spark. Unlike Anthropic's Claude Mythos Preview announcement yesterday, Muse Spark is actually available publicly today. Meta is also taking on the top Gemini and ChatGPT models with a new Contemplating mode that it says will roll out gradually. Last year, Meta retooled its AI efforts under the leadership of Alexandr Wang with a new team called Meta Superintelligence Labs. Today, Meta is releasing Muse Spark, which it calls "the first step on our scaling ladder" as the team builds a family of Muse models. "Muse Spark offers competitive performance in multimodal perception, reasoning, health, and agentic tasks," Meta says. "We continue to invest in areas with current performance gaps, such as long-horizon agentic systems and coding workflows." As part of the announcement, Meta shows how Muse Spark Thinking benchmarks favorably next to Anthropic's Claude Opus 4.6 Max, Google's Gemini 3.1 Pro High, OpenAI's GPT-5.4 Xhigh, and xAI's Grok 4.2 Reasoning models. Muse Spark is available starting today in Meta AI apps. Muse Spark Thinking is just one mode. Muse Spark Contemplating mode will compete with Gemini 3.1 Deep Think and GPT-5.4 Pro. We're also releasing Contemplating mode, which orchestrates multiple agents that reason in parallel. This allows Muse Spark to compete with the extreme reasoning modes of frontier models such as Gemini Deep Think and GPT Pro. Contemplating mode provides significant capability improvements in challenging tasks, achieving 58% in Humanity's Last Exam and 38% in FrontierScience Research. Meta says it's also opening a private API preview for select partners. What is Muse Spark best for? Meta highlights multimodal expertise, specifically with "visual STEM questions, entity recognition, and localization." Meanwhile, health is another area of strength for Muse Spark, according to Meta: One major application of personal superintelligence is to help people learn about and improve their health. To improve Muse Spark's health reasoning capabilities, we collaborated with over 1,000 physicians to curate training data that enables more factual and comprehensive responses. Muse Spark can generate interactive displays that unpack and explain health information such as the nutritional content of various foods or muscles activated during exercise. You can learn more about the new Muse Spark AI models from Meta Superintelligence Labs here.

Introducing Muse Spark: Scaling Towards Personal Superintelligence

Today, we're excited to introduce Muse Spark, the first in the Muse family of models developed by Meta Superintelligence Labs. Muse Spark is a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration. Muse Spark is the first step on our scaling ladder and the first product of a ground-up overhaul of our AI efforts. To support further scaling, we are making strategic investments across the entire stack -- from research and model training to infrastructure, including the Hyperion data center. In this post, we'll first explore Muse Spark's new capabilities and applications. After these results, we'll look behind the curtain at the scaling axes driving our progress toward personal superintelligence.

Meta's AI Outlook Brightens With Muse Spark Release - Meta Platforms (NASDAQ:META)

What Is Muse Spark Muse Spark is another type of AI model, similar to ChatGPT and Gemini, that is designed to understand and generate information. It is built to integrate across Meta's other main products, including Facebook, Instagram, and WhatsApp, enabling AI assistants, recommendations, and automation. Market Reaction and Challenges Ahead Outlook Benzinga Disclaimer: This article is from an unpaid external contributor. It does not represent Benzinga's reporting and has not been edited for content or accuracy. Market News and Data brought to you by Benzinga APIs To add Benzinga News as your preferred source on Google, click here.

Meta Muse Spark on WhatsApp: What It Means for India

Meta launched Muse Spark on April 8, 2026, confirming in a blog post that it will replace existing Llama models on WhatsApp, Instagram, Facebook, and Messenger in the coming weeks. The features are currently rolling out in the US first. A Meta spokesperson told MediaNama that the model is currently only available on the Meta AI app and meta.ai and declined to comment on the WhatsApp rollout timeline. What this could mean for Indian users: A safety flag worth noting: Apollo Research, a third-party evaluator, found that Muse Spark had the highest rate of evaluation awareness among all models it tested, meaning the model detected when it was undergoing safety testing and adjusted its behaviour accordingly. This raises the question of whether it behaves the same way with real users as it does during safety checks. Meta acknowledged that this "may affect model behaviour on a small subset of evaluations" but called it "not a blocking concern for release." With no independent AI auditing requirement in India, there is no mechanism to verify whether the model behaves consistently in real-world use versus during testing. What Muse Spark is: Meta's superintelligence team, which Reuters reported was built following a $14.3 billion deal with Scale AI, and its CEO, Alex Wang, developed Muse Spark. The model is currently only on Meta's AI app and website. Independent evaluations show it matches Google, OpenAI and Anthropic in language and visual understanding but lags in coding and complex reasoning, gaps Meta acknowledges. It is not open source, unlike previous Llama models. Meta has shared only a "private preview" with unnamed partners, has not disclosed model size, and says it hopes to open-source future versions.