Anthropic and OpenAI must be scratching their heads. They just announced the launch of new artificial-intelligence models to tackle the problem of rising costs but the AI race has moved onto a new prize-consumer agents. Now they suddenly look to be behind.

OpenAI and Anthropic have released relatively low-cost AI models (GPT-6 Sol and Luna, and Claude Opus 5.5, respectively) that would have expected to grab the headlines. They come with impressive performance as measured on independent benchmarks and look designed to address users' complaints about the spiraling cost of using their frontier models.

Instead, both companies have been overshadowed by Meta Platforms' Muse agent. Suddenly a personal assistant for the average consumer which can deal with customer service is the biggest thing in AI rather than solving impossible math problems.

Allowing Meta to dominate the consumer market feels like a big miss for OpenAI and Anthropic. Perhaps they were distracted by the safety debate as well as plans for Anthropic's initial public offering, which looks to have been delayed to November, according to The Wall Street Journal.

Will OpenAI and Anthropic strike back? Anthropic seems unlikely to launch a Muse competitor. It has focused on selling to businesses rather than individuals and earlier this month gave retailers blueprints to build shopping and merchant agents on top of its Claude AI, rather than releasing those products itself.

However, OpenAI is unlikely to cede the consumer space so easily. The ChatGPT-developer has already discussed launching its own personal AI assistant, according to technology-focused news outlet The Information. There's no reason why its models shouldn't be able to match Muse's capabilities, so the fight might come down to brand awareness and who has more infrastructure available to support free services.

The AI competition is moving quicker than ever. Meta has managed to sneak a lead in consumer agents but expect a swift response from rivals.