Meta launches Muse Spark, a multimodal chatbot built into its apps
Key Points
- Meta launches Muse Spark, its first AI model from Superintelligence Labs, as the default chatbot in the Meta AI app.
- Muse Spark supports text, image, video and audio generation, handling complex constraints like avoiding specific letters.
- The model can pull live social data, simulate roles and produce polished graphics, showcasing social fluency.
- Coding assistance is built in; Muse Spark writes and explains code in multiple programming languages.
- Initial rollout covers the U.S.; broader deployment to Instagram, Facebook, Messenger, WhatsApp and AI glasses is planned.
Meta unveiled Muse Spark, its first AI model from the Superintelligence Labs, as the new default experience across the Meta AI app and website in the United States. The multimodal chatbot can generate text, images, video and even audio, handling complex constraints while staying socially fluent. Meta says the model prioritizes people and will soon roll out to Instagram, Facebook, Messenger, WhatsApp and upcoming AI glasses, positioning the service as a direct competitor to ChatGPT and Google Gemini inside the platforms billions already use.
Meta introduced Muse Spark this week as the centerpiece of its AI push, replacing a patchwork of feature‑by‑feature tools with a single, purpose‑built chatbot. The model, described by the company as the first product of its Superintelligence Labs, now powers the Meta AI app and the meta.ai website for U.S. users. A broader rollout to Instagram, Facebook, Messenger, WhatsApp and the company’s upcoming AI glasses is slated for later this year.
Unlike earlier Meta AI experiments that focused on narrow tasks—sticker creation, caption suggestions, or simple text generation—Muse Spark offers a full multimodal experience. Users can upload files, generate images, produce short videos and even create audio clips, capabilities that rival those of ChatGPT and Google Gemini. In a hands‑on test, the model wrote a sad song from a rubber duck’s perspective while deliberately avoiding the letter “e,” then rendered an audio track of the duck’s performance. The result was quirky yet coherent, demonstrating the model’s ability to follow complex constraints without breaking.
Meta positions Muse Spark as “purpose‑built for Meta’s products,” emphasizing its socially fluent personality. The chatbot can pull live data from the company’s platforms, simulate roles and respond in character. When prompted to act as a “Bigfoot tech‑support agent” and reference recent Facebook chatter about sightings, the model produced a plausible dialogue and paired it with a polished image. Such interactions suggest the system can blend real‑time social insight with creative output.
Developers also benefit from Muse Spark’s coding abilities. The model can write, debug and explain code in languages ranging from Python to C++. In a test, it generated a Python timer script after admitting it could not run background tasks itself. This mirrors the growing trend of AI assistants serving as pair‑programmers for both novices and seasoned engineers.
Meta’s rollout strategy hinges on embedding the chatbot directly into the apps users already frequent. Rather than directing traffic to a separate web portal, the company aims to make AI feel like a natural extension of scrolling, messaging and sharing. If successful, Muse Spark could capture a sizable share of the chatbot market without having to out‑perform rivals on pure conversational metrics.
The company has not disclosed pricing or subscription details, but the immediate availability of the model in the Meta AI app signals a shift toward broader consumer access. As the rollout expands, users can expect a more integrated AI experience across the Meta ecosystem, potentially reshaping how people create and consume content online.