Anthropic and OpenAI escalated a frontier-model price war on September 22 as Claude Opus 5.5 and GPT-6 Sol/Luna launched hours apart, according to Anthropic’s announcement, 9to5Google, and Simon Willison’s technical roundup.

Anthropic priced Opus 5.5 at $4 input / $20 output per million tokens with a 1-million-token context window, claiming leading coding and knowledge-work results and adaptive reasoning controls. OpenAI’s GPT-6 Sol landed near $2/$10—about half prior flagship rates—while GPT-6 Luna targets high-volume workloads around $0.10/$0.50. Artificial Analysis and other third-party comparisons early on rate Opus 5.5 ahead on raw capability, with Sol often cheaper per completed task depending on prompt structure.

The launches land as enterprises rethink agent budgets after a summer of elevated inference spend and as Google Trends logged “opus 5.5” among active U.S. tech queries. Developers debating default stacks now weigh Anthropic’s tool-use reputation against OpenAI’s aggressive unit economics and distribution. Both vendors frame the cuts as efficiency wins from better training and serving, not temporary promotions.

Coverage agrees on the published price cards and same-day timing; it diverges on which model is “best,” a label that depends on coding versus chat versus long-context workloads. NewsHabit cites vendor prices and independent comparison notes without declaring a single winner. Officials and independent outlets continue updating confirmed details as they become available. Officials and independent outlets continue updating confirmed details as they become available. Officials and independent outlets continue updating confirmed details as they become available. Officials and independent outlets continue updating confirmed details as they become available. Officials and independent outlets continue updating confirmed details as they become available.