Anthropic follows OpenAI with frontier model price cuts

Is a frontier model price war brewing? Anthropic announced it is keeping the introductory prices for Claude Sonnet 5, less than two weeks after OpenAI cut prices for two of its top models.

Anthropic said access to Sonnet 5 would continue to cost $2 per million input tokens and $10 per million output, as it did at launch in June, as the company scrapped plans to introduce a $3/$15 structure in September.

While Anthropic did not explain the decision, the costs of frontier AI models have come under increased scrutiny as companies manage soaring AI budgets and competition improves from free, open-weight alternatives.

It did not return The Stack’s request for comment before publication.

On July 30, OpenAI cut the cost of its GPT-5.6 Luna model, the cheapest of its GPT-5.6 family, by 80% and the mid-tier GPT-5.6 Terra model by 20%, tying the move to under-the-hood efficiency gains.

A release said: “These updates help customers get more from every dollar they invest in AI and move faster when time matters.” The new prices are $2 per million tokens input and $12 per million output for Terra, and $0.20/$1.2 for Luna.

Model costs rise