Consumers are paying nearly 25% less for artificial intelligence models from leading U.S. labs than they were in mid-July, following price cuts announced by OpenAI and Anthropic, the Financial Times reported Friday (Aug. 14), citing data from Silicon Data.
OpenAI cut the price of its mid-tier model, GPT-5.6 Luna, by 80%, while Anthropic launched its Claude Opus 5 at half the price of its Fable 5, according to the report.
These price cuts came at a time when companies are becoming more cost-conscious after seeing their AI bills rise, and when Chinese developers such as Moonshot and DeepSeek are gaining ground among U.S. and European customers by offering models that are cheaper and increasingly closer to the capabilities of leading U.S. models, the report said.
The report said that while OpenAI and Anthropic have lowered the prices of their mid-tier models, they’re maintaining or increasing the prices of their top-tier models. It added that a more capable, more expensive model may be able to complete a task with fewer tokens and therefore at a lower cost.
PYMNTS has reported over the last two months that the AI price war is real and has reached American consumers, that soaring AI costs have pushed enterprise buyers to cheaper Chinese models and that the era of “tokenmaxxing,” or pushing employees toward the biggest AI models and the heaviest usage, is ending after two years of unchecked growth.
When OpenAI announced July 30 that it was cutting prices on select models, the company said in a blog post that it made the changes to improve the models’ performance per dollar across enterprise workloads.
On that day, OpenAI cut the price of GPT-5.6 Luna by 80%, cut the price of GPT-5.6 Terra by 20% and provided faster performance of GPT-5.6 Sol in the API while leaving its price unchanged. GPT-5.6 Sol is the company’s frontier model.
“We are building a resilient infrastructure portfolio and matching each workload to the systems best suited to run it,” OpenAI said in the post. “That approach supports both ends of the price-performance curve.”
It was reported Thursday (Aug. 13) that DeepSeek is adding peak-hour pricing for its flagship V4 models that will quadruple the current levels, though the new prices remain lower than those of its main competitors.