OpenAI and Anthropic Cut AI Costs

According to Ars Technica, the frontier AI race has shifted toward comparison shopping. Both OpenAI and Anthropic recently launched new models designed primarily to slash operational expenses for enterprise customers rather than introduce groundbreaking capabilities.

How much do the new Anthropic models cost?

Anthropic launched Opus 5.5, aiming to make its core workhorse model significantly cheaper. Input and output tokens now cost $4 and $20 per million respectively, representing a 20% price drop compared to Opus 5. Furthermore, cache reads dropped to $0.20 per million tokens—a 60% reduction.

What changes with OpenAI's GPT-6 Sol and Luna?

OpenAI introduced GPT-6 Sol and Luna as efficient, budget-friendly alternatives to its heavy-duty Astra model. According to Ars Technica, these models cost about half as much as their predecessors while delivering minor performance gains. Sol’s API pricing sits at $2 per million input tokens and $10 for output, while Luna charges $0.10 and $0.50.

What is the practical consequence for development teams?

For engineering teams running high-volume agentic workflows, these price cuts change deployment economics. Anthropic claims that Opus 5.5 reduces typical workload expenses by nearly 40% because it requires fewer total tokens to complete tasks, alongside a 30% increase in output generation speed.