Anthropic and OpenAI release lower-cost AI models

Anthropic and OpenAI have released lower-cost AI models, escalating a contest that is increasingly about making powerful systems affordable enough for everyday business use. Anthropic says its new Claude Opus 5.5 is faster and cheaper to run than before, while delivering its strongest results yet in the company’s own safety testing. That matters because companies pay for AI largely by how much computing each task consumes, so lower prices can turn tools for coding, research, and customer work from an experiment into a routine expense. Anthropic’s safety result is a company claim rather than an independent verdict, but the paired launches underline the market’s new pressure: leading AI makers must improve capability without keeping premium performance reserved for the most expensive models.