OpenAI Slashes GPT-5.6 Prices for Luna and Terra

OpenAI cut GPT-5.6 pricing on Thursday, dropping Luna by about 80% to $0.20 per million input tokens and $1.20 per million output, and trimming Terra by 20% to $2 input and $12 output; Sol stays the same.

OpenAI reduced prices for two GPT-5.6 models on Thursday, lowering Luna by about 80% to $0.20 per million input tokens and $1.20 per million output tokens, and cutting Terra by 20% to $2 per million input tokens and $12 per million output tokens. Pricing for the top-tier Sol model is unchanged.
The company attributed the cuts to efficiency gains in the systems that serve the models. The update arrives roughly three weeks after the GPT-5.6 launch, earlier than typical pricing adjustments.
Luna is the fastest and lowest-cost variant of GPT-5.6 and now carries the steepest discount. Terra is positioned as the mid-range option. Sol, the most powerful version in the series, keeps current rates.
Prices are set per million tokens for both input and output. Tokens measure the text the model reads and generates. Newer reasoning models often use more tokens during longer, agent-like tasks, prompting providers to refine pricing, rate limits and usage policies.
Lower-cost open-weight models from Chinese developers have added price pressure in the market. Providers have adjusted pricing and usage policies as customers focus more on costs while testing larger workloads.
OpenAI presented the reductions as part of efforts to deliver more capability per dollar and make higher-capacity models more cost-effective for developers and enterprises, while holding Sol steady. The pricing shift also comes as Big Tech earnings put AI spending, data-center costs and profitability under closer investor scrutiny.
