Llama is out; Muse Spark has entered. Meta has launched its renewed AI efforts with a new model it calls Muse Spark.
Unlike Anthropic’s Claude Mythos preview announcement yesterday, the Muse Spark is actually publicly available today.
The Meta also takes on the top-of-the-line Gemini and ChatGPT models with a new Contemplating mode that will roll out gradually.
The Muse Spark replaces the Llama models
Last year, Meta renewed its AI efforts under the leadership of Alexandr Wang with a new group called Meta Superintelligence Labs. Today, Meta is releasing the Muse Spark, calling it “the first step in our scaling process” as the team builds the Muse family of models.
“Muse Spark provides competitive performance in multimodal perception, reasoning, health, and agent functions,” Meta said. “We continue to invest in areas with current operational gaps, such as long-term agency systems and coding workflows.”
As part of the announcement, Meta shows how the Muse Spark Thinking benchmarks stack up against 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 for Meta AI apps.
Thinking mode is coming
Muse Spark Thinking is just one way. Muse Spark Contemplating mode will compete with Gemini 3.1 Deep Think and GPT-5.4 Pro.
We also release an inference mode, which organizes multiple agents that think in parallel. This allows the Muse Spark to compete with the extreme thinking methods of borderline models such as Gemini Deep Think and GPT Pro. Thinking mode provides a significant improvement in performance 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 to select partners.
Health is one area of Spark’s expertise
What is the best Muse Spark? Meta highlights multimodal technology, especially for “virtual STEM questions, business recognition, and localization.”
Meanwhile, health is another area of Muse Spark’s power, according to Meta:
One major application of personal intelligence is to help people learn and improve their lives. To develop Muse Spark’s health imaging capabilities, we’ve partnered with more than 1,000 physicians to curate training data that powers authentic and comprehensive answers. Muse Spark can generate interactive displays that unpack and explain health information such as the nutritional content of various foods or the muscles activated during exercise.
You can read more about the new Muse Spark AI models from Meta Superintelligence Labs here.
Meanwhile, we should finally see the results of the new agreement between Apple and Google in less than two months at WWDC 2026. iOS 27 with features Gemini-powered Siri and Apple Intelligence will be revealed on June 8.
FTC: We use auto affiliate links to earn income. More.
