OpenAI Cuts AI Model Prices as Businesses Question Rising Costs

OpenAI slashed prices on two of its AI models this week. The move comes as companies push back on high AI bills and cheaper Chinese rivals grab market share.

What Changed

On Thursday, OpenAI cut the price of its smaller GPT-5.6 Luna model by 80%. It also trimmed the mid-tier GPT-5.6 Terra model by 20%. The company left its flagship Sol model untouched.

Both discounted models launched only three weeks ago. That makes this an unusually fast price cut.

Here is the new cost breakdown, measured in “tokens,” the units companies use to track AI usage:

• Luna (input): 20 cents per million tokens, down from $1
• Luna (output): $1.20 per million tokens, down from $6
• Terra (input): $2 per million tokens, down from $2.50
• Terra (output): $12 per million tokens, down from $15

OpenAI says efficiency gains in its newer models made the cuts possible. Smaller, cheaper models can now handle tasks that once needed a top-tier system.

Why Now

Businesses are growing wary of ballooning AI costs. Many are shifting from flat subscription plans to usage-based billing. That shift can lead to unpredictable monthly bills, and companies want more control over their spending.

Chinese competitors add more pressure. Z.ai’s GLM-5.2 model nearly matches the performance of top U.S. models at a lower price. That kind of competition pushes American labs to rethink how they charge customers.

Tech leaders across the industry have said cheaper AI is key to wider adoption. Thursday’s price cut backs up that view.

The Anthropic Angle

OpenAI’s move also puts pressure on Anthropic. Anthropic’s Claude models lead in enterprise and developer use, but they sit on the pricier end of the market.

The numbers make the gap clear. OpenAI’s Terra model now costs less than Anthropic’s mid-tier Claude Sonnet 4.6 model, which charges $3 per million input tokens. That makes Terra a cheaper option for many of the same tasks.

What Analysts Are Watching

Analysts expect the price cuts to boost usage of both OpenAI’s and Anthropic’s tools. Lower prices tend to pull in more customers and more workloads.

But there’s a catch. These cuts could squeeze profit margins at a sensitive time. Both OpenAI and Anthropic are pouring billions into computing infrastructure, and both face highly anticipated IPOs down the road. Thinner margins now could complicate those plans.

For businesses, the takeaway is simple: AI costs are shifting fast, and the cheaper options just got cheaper. Companies that track their AI spend closely may want to revisit their model choices soon.