OpenAI reduced the usage prices of its GPT-5.6 Luna and Terra models on July 30, 2026. According to the company's official announcement, the largest reduction applies to Luna, the model positioned for fast and lower-cost workloads. Luna's API input and output prices were cut by 80%, while prices for Terra, the balanced tier for everyday work, were reduced by 20%. OpenAI announced no price change for Sol, the highest-capability model in the GPT-5.6 family.
What are the new API prices?
Under the updated schedule, GPT-5.6 Luna costs $0.20 per million input tokens and $1.20 per million output tokens. Terra now costs $2 per million input tokens and $12 per million output tokens. When OpenAI introduced the GPT-5.6 family, Luna was priced at $1 for input and $6 for output, while Terra cost $2.50 for input and $15 for output. The published old and new prices therefore directly support the announced reduction percentages.
What changes for ChatGPT and Codex users?
The official announcement says the adjustment is not limited to API billing. Luna and Terra usage will consume fewer credits under paid subscriptions when people use the models through ChatGPT Work and Codex. However, OpenAI says the listed subscription prices and the total quota budgets assigned to subscribers will remain unchanged. The update therefore does not represent a direct reduction in the subscription fee. Instead, it means eligible users can perform more Luna or Terra work within the same credit budget.
How does OpenAI explain the reduction?
OpenAI attributes the new prices to efficiency improvements in the systems serving the models. The company says the change is intended to make Luna more economical for high-volume work. Reuters reported that the move arrives as businesses examine artificial intelligence spending more closely and competition among model providers increases. Axios also noted that the cuts were announced about three weeks after the broader launch of the GPT-5.6 family, making the timing unusually close to the original release.
Sol pricing and the cloud rollout
Sol pricing remains unchanged under the announcement. OpenAI also said the Luna and Terra prices take effect from July 30, while the corresponding changes on AWS would begin rolling out later the same day. Organizations using different service channels should check the rate shown in their own accounts and review their usage records, because the precise timing of availability may vary while a staged rollout is underway.
What the change means for developers
The new schedule can lower unit costs for applications that process a high number of short requests, including classification, routine data processing, and automated business workflows. Actual spending still depends on more than the advertised token rate. Input and output length, reasoning-token use, prompt caching, and the selected service tier can all affect a final bill. Developers should therefore update budgets prepared with the previous prices and compare the new estimates with measured usage from their own systems.
The verified changes at a glance
The central figures are consistent across the official announcement and independent reporting. Luna input pricing fell from $1 to $0.20 per million tokens, and output pricing fell from $6 to $1.20. Terra input pricing moved from $2.50 to $2, while output pricing declined from $15 to $12. Sol pricing and total subscription quota budgets did not change. The announcement concerns the cost of using existing GPT-5.6 tiers rather than a new model version or a newly announced capability.
