Meta and Google Advance AI Capabilities While Rivals Cut Costs

Meta and Google have significantly advanced their artificial intelligence capabilities this week, focusing on high-performance frontier models. Meta released Muse Wonder and Muse Wonder Full, which are designed specifically for AI agents. Simultaneously, Google made Gemini 4 Argon available to the public, noting that the model achieved a top score on tasks related to finance, legal matters and taxation. These moves represent a push to raise the ceiling on what the most powerful AI systems can achieve.

In contrast, Microsoft and Cloudflare have focused on reducing the cost and complexity of AI operations. Both companies launched decision models that select from fixed options rather than generating free-form text. Microsoft stated that its new model operates 35 times faster than OpenAI’s GPT-6 Sol and cost 200 times less in one internal test. This approach allows for quick yes-or-no and multiple-choice decisions to be made for a fraction of a cent, targeting routine operational tasks.

The broader AI market is seeing a divergence in strategy, with major tech firms pushing towards more powerful but expensive frontier models while others optimise for efficiency. Smaller developers, including Liquid AI, JetBrains and Noiz AI, have also contributed to this trend by releasing open models for on-device agents, coding agents and real-time video with sound. As the frontier becomes more crowded and costly, enterprise spending is increasingly focused on matching specific tasks to the cheapest capable model. AT&T has already adopted this strategy, reporting that it cut the cost of coding and other advanced AI tasks by up to 56% by routing queries to cheaper models.