In a 24-hour period, OpenAI, SpaceXAI, and Meta turned AI into a race to the bottom on price. On Wednesday, Elon Musk’s SpaceXAI launched Grok 4.5, its first model since the company went public. A day later, OpenAI moved GPT-5.6 to general availability, with a model family arrangement similar to Anthropic’s: Sol at the top, Terra in the middle, Luna at the bottom. Meta shipped its first-ever paid model in the same window. While the models differ in capabilities, the shared argument in all launch copy was price. OpenAI sold GPT-5.6 on “more intelligence from every token” and bragged that Luna beats Anthropic’s Opus 4.6 at roughly a quarter of the cost. Musk pitched Grok as Opus-class, “but faster, more token-efficient and lower cost.” Sam Altman went on CNBC and led with a number: Sol is 54% more token-efficient on agentic coding, because “every enterprise now is thinking about spend.” The one breakthrough every lab put in its headline this week was the bill, responding to recent headlines that AI might not be providing a significant enough return on investment.