OpenAI Just Cut Its New AI Prices in Half — Here's Why You Care

OpenAI just put two new AI models on what amounts to a half-price sale. They're called GPT-6 Sol and GPT-6 Luna, and they follow the bigger, fancier GPT-6 Astra released earlier this month. Sol is the smart one — built for hard work like writing software and running long, multi-step tasks. Luna is the cheap, fast one — built for high-volume chores like summarizing thousands of documents. Both cost roughly half what OpenAI charged for the previous generation.

Why does a price cut matter to you? Because AI agents — think of them as digital assistants that can actually take actions, not just chat — get expensive fast. Every time an agent looks something up, checks its work, or tries another step, it racks up a bill. When that bill is high, companies limit how much their AI is allowed to do. When it drops, they let it run longer and try harder. OpenAI says one of its new models finishes a complex task for 27 cents, while a rival model doing similar work costs about eleven times more.

There's a privacy and reliability angle too. These models can chew through huge amounts of information quickly and cheaply, which makes it more practical for companies to feed your emails, documents, or records into them. OpenAI says Sol makes about half as many factual mistakes as its predecessor, and both models reuse information they've already seen instead of reprocessing it. That saves money and energy.

The catch: all these performance numbers come from OpenAI's own testing, not independent reviewers. Real-world results can differ. And these are developer prices — the savings reach you indirectly, through cheaper apps and tools, not as a discount you buy directly. If an AI product you use suddenly gets smarter or cheaper, this is likely part of the reason.