Key facts
- OpenAI has cut prices on its GPT-5.6 Luna and Terra AI models.
- The GPT-5.6 Luna model's price was reduced by 80%, and the Terra model by 20%.
- The price of OpenAI's flagship Sol model remains unchanged.
- The new pricing aims to attract businesses concerned about AI costs and compete with rivals.
- OpenAI's smaller models can now perform tasks that previously required higher-tier systems.
OpenAI has significantly reduced the prices of its lower-tier AI models, GPT-5.6 Luna and Terra, by 80% and 20% respectively, while maintaining the price of its flagship Sol model. This strategic move comes as businesses increasingly scrutinize their artificial intelligence expenditures, facing ballooning costs and seeking more affordable solutions.
The price cuts are expected to intensify competition within the AI industry, particularly against cheaper Chinese open-source rivals like Z.ai's GLM-5.2, which offer comparable performance at a lower cost. OpenAI's decision also puts pressure on competitors such as Anthropic, whose Claude models are popular in enterprise but are positioned at a higher price point.
Analysts suggest that while these price reductions could boost usage of OpenAI's and Anthropic's technologies, they might also strain their financial resources ahead of anticipated initial public offerings. The new pricing structure means that businesses will pay less per million tokens, with Luna's input and output costs dropping to $0.20 and $1.20, and Terra's to $2 and $12, respectively. This makes OpenAI's smaller models capable of performing tasks that previously required its top-tier systems.
OpenAI attributes the price reductions to efficiency gains derived from its GPT-5.6 model, which has improved code and optimized performance. Although token prices have been declining over the past year, the overall cost of completing tasks has been rising due to a shift from flat subscriptions to usage-based pricing, leading to unpredictable and often higher bills for companies.