OpenAI Cuts Prices and Advances Model Efficiency

OpenAI has unveiled a strategic direction centered on delivering "abundant intelligence"—the vision that true value emerges when AI becomes simultaneously more powerful, more accessible, and more affordable. The company contends that as the cost of useful intelligence declines, it enables a broader range of practical applications and business opportunities. The organization announced substantial price reductions across GPT-5.6 models. Luna dropped by 80%, now priced at $0.20 per million input tokens and $1.20 per output tokens. Terra received a 20% reduction, while Sol introduced a Fast mode delivering 2.5x faster inference at double the standard cost. These adjustments give customers granular control over the tradeoffs between intelligence quality, processing speed, and expenditure. Behind the pricing changes, OpenAI detailed engineering gains improving operational efficiency. Model-serving costs fell by 20% while token-generation efficiency improved more than 15%. Enhancements to context handling and reasoning mechanisms lifted benchmark performance substantially, particularly on the public ARC-AGI-3 task set, while simultaneously reducing token consumption. The platform now serves over one billion active users and two million businesses. Agentic work has become dominant, constituting 99.8% of weekly output tokens, reflecting a shift toward autonomous systems handling complex, multi-step tasks across enterprise operations.