OpenAI made a bold move by cutting the price of its GPT-5.6 Luna model by 80 percent overnight, shaking up the AI pricing landscape. At the same time, Terra’s cost dropped by 20 percent. But the surprises didn’t stop there. The blazing-fast Sol model didn’t see a price cut, yet it got an upgrade with a new Fast mode that accelerates processing speeds up to 2.5 times without any loss in output quality.

Massive Cost Reductions on Luna and Terra Models

From July 30, developers will pay just $0.20 for every million input tokens and $1.20 for output tokens using Luna, making it the cheapest and quickest option in the GPT-5.6 lineup. Terra users gain a 20 percent price cut, now paying $2 per million input and $12 per million output tokens. OpenAI credits these savings to extensive improvements in model efficiency, inference infrastructure, and software optimization.

Sol Model Gains Speed Without a Price Drop

While Sol’s base pricing remains unchanged, OpenAI replaced its Priority Processing with Fast mode. This new tier runs up to 2.5 times faster than standard processing, with a price set at twice the standard rate $10 per million input tokens and $60 per million output tokens. Existing API users on the priority tier will see a smooth transition, as the upgrade is backward compatible and requires no code changes.

OpenAI highlights that Luna now delivers performance comparable to top-tier models from just a year ago but at nearly nine times the speed and at a cost efficiency of about 6 cents per dollar spent per task. The Luna model has also beaten competitors like Fable 5 on AI exams while slashing costs by almost 99 percent.

Efficiency Gains Driven by AI Itself

Interestingly, part of these breakthroughs comes from the Sol model’s own involvement in optimizing its internal processes. Under human guidance, Sol rewrote production kernels, ran experiments to enhance token generation, and improved training supervision. These efforts led to kernel optimizations that contribute significantly to the overall performance boost.

These changes make it more affordable for businesses running large-scale AI workloads such as document analysis, customer service classification, and automated coding to scale without breaking the bank.

This article is for informational purposes only and does not constitute financial advice.