OpenAI reduces the price of GPT-5.6 by up to 80%.

OpenAI has announced a price reduction for its GPT-5.6 series.

Advancing the price-performance frontier with GPT-5.6 | OpenAI

We are committed to pushing the model frontier across cost efficiency, capability, and speed.

Starting today, we are reducing prices for GPT-5.6 Luna by 80% and GPT-5.6 Terra by 20% , and offering a faster option for GPT-5.6 Sol in the API.

Luna and Terra's lower prices are… pic.twitter.com/rFhK7XKedp

— OpenAI (@OpenAI) July 30, 2026

The GPT-5.6 series is offered in three types: the top-performing 'GPT-5.6 Sol,' the balanced 'GPT-5.6 Terra,' and the lower-priced 'GPT-5.6 Luna.' The price reductions announced this time apply to the GPT-5.6 Terra and GPT-5.6 Luna.

At the time of release, the price per 1 million tokens was $2.50 (approximately 401 yen) for input and $15 (approximately 2407 yen) for output for GPT-5.6 Terra, and $1 (approximately 161 yen) for input and $6 (approximately 963 yen) for output for GPT-5.6 Luna. With this price reduction, the prices for GPT-5.6 Terra are now $2.00 (approximately 321 yen) for input and $12.00 (approximately 1926 yen) for output, and for GPT-5.6 Luna, they are $0.20 (approximately 32 yen) for input and $1.20 (approximately 193 yen) for output. This represents a 20% discount for GPT-5.6 Terra and an 80% discount for GPT-5.6 Luna.

The graph below shows the cost-effectiveness of various AI models, with the horizontal axis representing the cost per task and the vertical axis representing intelligence performance. With this price reduction, the GPT-5.6 Luna has become more cost-effective than not only closed models such as the Claude Sonnet 5 and Gemini 3.6 Flash, but also Chinese-made open models such as the DeepSeek V4 Pro and GLM-5.2.

In addition to the price reduction, a 'high-speed mode' has been introduced to the API. High-speed mode enables faster processing than the standard mode while maintaining the same performance. Using high-speed mode with GPT-5.6 Sol doubles the price but increases processing speed by approximately 2.5 times.