OpenAI slashed prices on its smaller AI models on Thursday, cutting the cost of its low-tier GPT-5.6 Luna model by 80% and its mid-tier Terra by 20%, while leaving its flagship Sol unchanged. Sending text to Luna drops to 20 cents per million tokens from $1, and generating responses falls to $1.20 from $6. Terra's input price falls to $2 from $2.50, with output dropping to $12 from $15.
The cuts reflect mounting cost scrutiny from businesses facing hefty AI bills, and intensify competition as US labs battle cheaper open-source rivals from China that nearly match their performance at a lower cost. Anthropic's mid-tier Claude Sonnet 4.6, at $3 per million input tokens and $15 for output tokens, now sits above Terra's rates. Analysts say the price cuts could boost usage of OpenAI's models but strain the company's finances ahead of a highly anticipated IPO.
OpenAI said the reductions were partly enabled by efficiency gains from GPT-5.6, including the model's ability to improve code and optimize performance during internal development. While token prices have been falling over the past year, the cost of completing a task is rising as AI firms shift from flat subscriptions to usage-based pricing — leaving companies with unpredictable and often higher bills as usage per task becomes harder to estimate.
