OpenAI announced on Thursday that it is reducing the prices of its latest artificial intelligence models, GPT-5.6 Terra and GPT-5.6 Luna, just roughly three weeks after their public release.

The company is under pressure to cater to a more cost-conscious customer base, as enterprise clients are hesitant to deploy expensive models without clear returns on investment. At the same time, OpenAI must fend off competition from tech giants like Google and Microsoft, which have been promoting cost-effective models.

OpenAI launched three GPT-5.6 series models, including the most powerful Sol, the mid-range Terra, and the fastest Luna. The company stated that Terra has been reduced by 20%, now charging $2 per million input tokens and $12 per million output tokens. Luna has been cut by 80%, now charging $0.20 per million input tokens and $1.20 per million output tokens. The price of Sol remains unchanged.

As AI bills surge, many enterprises are actively controlling spending, a shift closely monitored by AI developers. Open-weight AI models, which allow users to download, modify, and run on their own infrastructure, offer a cost-effective alternative for businesses.

Competitor Anthropic subsequently launched a new model, Claude Opus 5, claiming it offers the best value for money. During an earnings call, Microsoft CEO Satya Nadella also repeatedly emphasized the cost-effectiveness of its models. Google similarly introduced three new models this month, aiming to compete on cost advantages.