OpenAI cuts GPT-5.6 prices by up to 80%

OpenAI has aggressively lowered the pricing for two of its recently launched GPT-5.6 models to stimulate enterprise adoption. Announced on July 30, 2026, the price cuts target the mid-tier GPT-5.6 Terra and the entry-level GPT-5.6 Luna, both of which debuted just three weeks earlier. The cost of Luna, OpenAI's fastest and cheapest model, has plummeted by 80 percent, bringing its API pricing down to $0.20 per million input tokens and $1.20 per million output tokens. Meanwhile, the everyday-work model, Terra, received a 20 percent discount, establishing its new rates at $2 per million input tokens and $12 per million output tokens.

While the flagship GPT-5.6 Sol model retains its original pricing, OpenAI has introduced a new Fast mode in its API to replace Priority Processing. This update reportedly allows Sol to run at speeds more than two-and-a-half times faster than standard processing. According to the company, these adjustments were made possible by efficiency breakthroughs achieved during the development of the GPT-5.6 foundation model.

These reductions position OpenAI aggressively against rivals like Anthropic, whose mid-tier Claude Sonnet 4.6 is priced higher at $3 per million input tokens and $15 per million output tokens. The price drop also comes at a time when businesses are scrutinizing their AI investments; for instance, Uber capped its per-employee AI spending in June. By lowering barriers to entry, OpenAI aims to maintain its edge over domestic competitors like Google and Microsoft, as well as Chinese firms such as Alibaba and Moonshot, particularly as both OpenAI and Anthropic prepare for potential public offerings.

For developers and enterprise architects, this pricing shift significantly alters the cost-benefit analysis of deploying generative AI at scale. The massive discount on Luna makes high-volume, low-latency tasks far more viable, while the cheaper Terra rates offer a direct discount on daily workflows. Practitioners can now reallocate budgets to run more extensive agentic pipelines or high-frequency queries without exceeding strict corporate spending limits.