
OpenAI has announced a major cost reduction for its GPT-5.6 model family, with the Luna variant now 80% cheaper. This price cut is attributed to the Sol model's ability to optimize its own GPU code, improving efficiency by 15% and reducing serving costs by 20%. The Luna model now offers a cost-effective solution at $0.20/$1.20 per million tokens, making it a leader in cost per task. OpenAI's move challenges competitors like Google's Gemini Flash, emphasizing the company's focus on providing affordable and intelligent AI solutions.
Read original