Meta Launches Muse Spark as Investors Seek Clarity

Meta Platforms unveiled Muse Spark in early April 2026, positioning it as the first model from its newly formed Meta Superintelligence Labs. The announcement represents a significant pivot from the company's previous open-source strategy with Llama models toward a closed-source, paid-access approach similar to OpenAI and Anthropic.

According to the official Meta announcement, the model is built on a ground-up rebuild of the AI stack completed over nine months. Muse Spark is designed as a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration. It now powers the Meta AI assistant in the Meta AI app and meta.ai.

The physical experience of using Muse Spark differs meaningfully from previous iterations. Users can snap a photo of an airport snack shelf and have the model identify and rank snacks by protein content without squinting at labels. The interface handles multiple subagents in parallel — one drafts an itinerary, another compares destinations, a third finds kid-friendly activities — all simultaneously. (This is the kind of parallel processing that has been promised for years, finally arriving in something resembling usable form.)

Performance benchmarks show mixed results. On Arena.AI, Meta AI trails Anthropic's Claude and Google's Gemini in text capabilities but only Claude in vision. It currently ranks ahead of OpenAI's GPT in both categories. The Artificial Analysis Intelligence Index gives Muse Spark a score of 52, placing it in the top five but still behind Gemini 3.1 Pro Preview, ChatGPT 5.4, and Claude Opus 4.6, which score between 53 and 57.

The improvement from Meta's previous models is stark. Llama 4 Maverick scored just 18 on the same index. On the Agentic Index, Meta jumped from seven with Maverick to 62 with Muse Spark. The gap on coding tasks remains wide, though.

Wall Street's reaction has been cautiously optimistic. Shares jumped more than 9% in the two days following the announcement, easily surpassing the S&P 500's 3% gain over the same period. CNBC reporting notes that analysts expect year-over-year revenue growth of 31% for the first quarter to $55.6 billion, representing the fastest expansion rate since 2021.

However, investors remain focused on monetization clarity. Citizens analysts described AI as a "complementary good" for Meta and said they expect to hear more on the company's earnings call. They're awaiting a strategy to drive scaled consumer usage akin to ChatGPT and Claude, which could unlock new data and ad budgets.

The leadership shift behind Muse Spark is substantial. Alexandr Wang, former CEO of Scale AI, now leads Meta Superintelligence Labs following Meta's $14.3 billion investment in the data-labeling startup. Mark Zuckerberg also brought in former GitHub CEO Nat Friedman and Daniel Gross, previously CEO of AI startup Safe Superintelligence.

Meta is simultaneously cutting headcount as it doubles down on AI. The company announced it would lay off 10% of its workforce, approximately 8,000 employees, on May 20 to improve business efficiencies. This happens as Meta projects 2026 AI-related capital expenditures in the range of $115 billion to $135 billion, up from $72.2 billion in 2025.

Loop Capital analysts wrote that Meta's hefty investments have fed a negative perception that it's "a company desperately spending to fix problematic AI initiatives." The release of Muse Spark shows Meta is producing AI models that could further improve its core online ad business. Foundational LLM/agentic reasoning models are certainly key, but image/video generation models may have greater near-term engagement and monetization implications.

Whether users actually pay for Muse Spark remains the real question. The model is available now at meta.ai and the Meta AI app, with a private API preview opening to select users. Future versions may be open-sourced, but the immediate strategy focuses on paid developer access.

The ad business continues to grow, boosted by increased targeting capabilities from AI advancements. Even if Muse Spark and future models fail to outperform rival systems, those tests are of mixed importance because of Meta's clear advantage in ads. The real bar for success is building models that power excellent products for users, creators, and advertisers.

Time will tell if the closed-source shift pays off. Meta's stock price is up 24% in the past year, while Alphabet shares have gained 116% over that stretch, boosted by Gemini growth. The company is back in the AI conversation, but whether that translates to sustained investor confidence depends on execution that hasn't happened yet.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn