OpenAI’s release of the GPT-5.4 model has garnered attention as it reaches a score of 51, equivalent to that of Luna, the company’s efficiency-focused model, which serves as a comparison for pricing and performance. OpenAI has priced GPT-5.4 significantly lower, with a current cost of $2.50 per full token compared to Luna’s $0.20. This pricing strategy aligns with OpenAI’s goal to compete effectively in a marketplace that increasingly prioritizes low inference costs without sacrificing intelligence, a move that reflects broader industry trends toward differentiated model offerings.

Luna: Luna is OpenAI’s GPT-5.6 Luna model built for cost-sensitive, high-volume applications. It delivers intelligence levels matching earlier flagship models while forming part of a tiered lineup that also includes higher-end options. The model targets users seeking speed and economy over maximum reasoning depth.
OpenAI: OpenAI develops and deploys frontier AI models including successive GPT generations. It recently released the GPT-5.6 family that includes Luna as its efficiency-focused tier. The launch directly addresses user demand for flagship performance at lower operational costs.
Nic Dunz: Nic Dunz is an AI analyst and commentator affiliated with Krea AI who tracks model capabilities and pricing dynamics. He posted a direct comparison of GPT-5.4 performance and cost against the newly available Luna model. His analysis illustrates the compression of AI economics occurring over recent months.

Model Family: OpenAI previewed the GPT-5.6 series comprising Sol for peak performance, Terra for balanced use, and Luna for efficiency-focused tasks.
Pricing Strategy: The introduction of Luna positions OpenAI to compete in a market increasingly sensitive to inference costs while maintaining high intelligence benchmarks.
Competitive Landscape: OpenAI’s tiered releases respond to broader industry moves toward differentiated model offerings that balance capability with accessibility.