The ChatGPT creator has slashed the price of its smaller GPT-5.6 Luna model by 80% and the mid-tier Terra model by 20%, while maintaining the price of its flagship Sol model.
This pricing adjustment reflects the growing scrutiny over costs by businesses dealing with substantial AI expenses, prompting American labs to reassess their pricing strategies. Many tech leaders have recently indicated that more affordable AI options are crucial for broader adoption of the technology.
OpenAI’s updated pricing also puts additional pressure on Anthropic, whose Claude models are prevalent in enterprise and developer settings but command higher prices. Both companies face challenges from open-source Chinese competitors like Z.ai’s GLM-5.2, which deliver comparable performance at lower costs.
Experts suggest that reducing prices could enhance the usage of OpenAI’s and Anthropic’s technologies, but it may also strain their financial health as they approach much-anticipated initial public offerings.
Although Thursday’s price reductions apply only to OpenAI’s smaller and mid-tier models, the company asserts that this will benefit businesses overall, as these models can now accomplish tasks that previously required a higher-tier system at a significantly lower cost.
The new pricing structure indicates that businesses leveraging OpenAI’s technology will incur lower costs for every million “tokens,” the units used to measure AI utilization, processed through these models.
Sending text to Luna decreases to 20 cents per million tokens from $1, and Terra’s costs drop to $2 from $2.50, while generating responses reduces to $1.20 and $12 from $6 and $15, respectively. In comparison, Anthropic’s mid-tier Claude Sonnet 4.6 model is priced at $3 per million input tokens and $15 per million output tokens, which is higher than Terra’s rates.
OpenAI noted that these lower prices were made possible partly by efficiency improvements from GPT-5.6, including enhancements in code generation and performance optimization during internal development.
Overall, token prices have been on a downward trend over the past year; however, the costs associated with completing tasks are rising as AI companies transition from fixed subscriptions to usage-based pricing models.
This shift is resulting in unpredictable and frequently higher bills for companies as estimating usage per task becomes increasingly challenging.
Also Read: SEBI clears ₹1,491 crore settlement in NSE co-location, dark fibre cases
