OpenAI cut pricing for its GPT-5.6 Luna model by 80% and Terra by 20%, passing efficiency gains from its latest foundation model directly to enterprise customers.

The price reductions take effect immediately across the API, with Luna becoming the company's most affordable frontier-class model at roughly 6 cents per task compared to similar capabilities from competitors.

Luna now handles high-volume workflows with tool use and multi-step reasoning at a fraction of previous costs. Terra, positioned as the balanced everyday model, delivers comparable quality to GPT-5.5 at half the cost per task and 40% faster processing, according to customer evaluations.

OpenAI also introduced Fast mode for its flagship GPT-5.6 Sol model, replacing Priority Processing with speeds up to 2.5x faster than standard processing at double the price.

Customer adoption accelerates

Early enterprise adopters report significant operational improvements. Replit described Luna as "intelligence too cheap to meter," while Notion found Terra delivered comparable quality to GPT-5.5 in 60% less time.

Cognition integrated Luna into its Devin Fusion product for cost savings without quality compromise. Dust reported 40% faster processing and 40% lower costs compared to their previous default model.

The pricing changes reflect what OpenAI calls a "compute strategy built for scale," matching workloads to optimal systems across its infrastructure portfolio.

GPT-5.6 Sol contributed to its own efficiency improvements by autonomously rewriting production kernels and running optimization experiments. The model reduced end-to-end serving costs by 20% and increased token-generation efficiency by over 15%.

OpenAI positions the updates as advancing its mission to make AGI benefits widely accessible. The company said businesses can now apply maximum useful intelligence at each workflow stage while paying appropriate prices for value created.

The new pricing structure allows enterprises to scale AI operations without sacrificing speed on frontier tasks, with Luna handling routine work and Sol managing complex reasoning when response time matters.