OpenAI cuts GPT-5.6 Luna pricing 80% and adds a faster Sol Fast API tier

OpenAI cut the price of its GPT-5.6 Luna model by 80% on July 30, to $0.20 per million input tokens and $1.20 per million output tokens, down from $1 and $6.

The reduction is the latest escalation in an AI price war that is shifting model competition toward cost. OpenAI attributed the cut to efficiency gains from GPT-5.6’s own code-optimization and token-generation improvements, roughly three weeks after the family launched.

GPT-5.6 Terra, the mid-tier model, dropped 20% to $2 and $12 per million input and output tokens, while pricing for the top-tier Sol model stayed at $5 and $30, OpenAI said.

The company also introduced Sol Fast, an API mode priced at $10 and $60 per million tokens that it says delivers roughly 2.5 times the throughput of standard Sol without changing the underlying model. Sol Fast replaces OpenAI’s earlier Priority Processing offering.

OpenAI’s throughput claim for Sol Fast has not been independently verified. The pricing math is straightforward enough: the new tier costs double standard Sol for buyers who want speed, while the Luna cut chases high-volume developers who care most about the per-token bill.

Aggressive price cuts on last-generation models have become a standard move as OpenAI, Google and Anthropic compete to keep developers on their platforms rather than defect to cheaper open-weight alternatives.