Meta's top artificial intelligence executive, Alexandr Wang, publicly jabbed at rival Google this week, asking, 'Gemini who?' after Meta's latest model, Muse Spark 1.3, posted a 62-point score on widely watched reasoning benchmarks — a result that placed it ahead of Google's Gemini family of models. The remark, circulated on social channels, underscored the increasingly combative tone among frontier-AI labs as they race to claim technical superiority.
Muse Spark 1.3's performance marks a sharp jump in Meta's reasoning capabilities. While exact benchmark components were not disclosed in the publicly available details, the score signals measurable gains over earlier Meta checkpoints and against competing systems from Google. Reasoning benchmarks, which test a model's ability to plan, infer, and solve multi-step problems, have become a key proxy for raw model intelligence, and a high score typically draws close attention from enterprise customers and developer communities.
ALSO READ | Apple’s First Foldable iPhone Duo Set for Tonight’s Launch with Leaked Color Lineup and Pricing
The exchange reflects the broader competitive landscape in generative AI, where companies including OpenAI, Anthropic, Google, and Meta trade barbs alongside benchmark disclosures. Wang, who leads Meta's AI division after his appointment following the company's multibillion-dollar investment in Scale AI, has positioned himself as a vocal critic of rivals, frequently weighing in on model launches from peers. His latest post immediately drew responses from researchers and industry observers tracking the ongoing model leaderboards.
For Google, the swipe lands at a sensitive moment. The Gemini line has been central to the search giant's strategy to integrate generative AI across Search, Workspace, and its cloud offerings. Any public perception of slippage in reasoning benchmarks can influence developer sentiment and enterprise procurement decisions. Analysts noted that Wang's tone, while rhetorical, points to real pressure on Google to demonstrate that Gemini's next iteration can close the gap or pull ahead again in the same evaluations.
