Google and Meta Rejoin the AI Race

Google and Meta have introduced new models they say can compete with offerings from OpenAI and Anthropic, reshaping a field that recently appeared dominated by those two companies.

Google released Gemini 4 Argon on Tuesday. The company says it matches or beats flagship OpenAI and Anthropic models on numerous benchmarks. During its introductory period, Argon costs up to 80 percent less per token than rivals, and remains significantly cheaper after the promotion ends.

Meta unveiled Muse Wonder Full on Monday. According to the company’s testing, it can compete on numerous benchmarks with leading models including GPT-6 Astra, Claude Opus 5.5, and Claude Fable 5.1. Meta had released Muse Spark in April as its first major model from Superintelligence Labs, but the source describes it as impressive rather than frontier-level.

The launches follow a period when Google’s Gemini 3, released late last year, struggled to keep pace with newer offerings. Google also saw changes to its AI teams and leadership. Meta, meanwhile, spent billions on AI talent and infrastructure, including Scale AI and its cofounder Alexandr Wang.

The article argues that Google and Meta’s large product ecosystems and user bases could give them advantages as competition shifts beyond model performance. It also notes rising safety concerns, rogue AI incidents, and improving open models as factors making it harder for any company to stay ahead. The claims about model performance and benchmarks are attributed to the companies.