OpenAI Enterprise Revenue Accelerates; Why Luna Claims to Be Cheaper Than Chinese Open-Source Models

OpenAI has long relied on consumer businesses such as ChatGPT subscriptions to build its revenue base, but enterprise clients are growing even faster.

Friar disclosed that the enterprise and consumer businesses were approaching parity by mid-year, reflecting that enterprise AI applications are moving from pilot stages toward larger-scale actual deployment. For OpenAI, this not only means more diversified revenue sources, but also indicates that its business model is becoming more deeply embedded in enterprise software, R&D, customer service, and automated workflows.

Enterprise clients care less about model capabilities themselves and more about whether a model can achieve a sustainable balance among cost, stability, and business return. This has also become the core battlefield in the competition between OpenAI and open-source models.

In terms of cost competition, Friar specifically highlighted OpenAI's low-cost model, Luna. She stated that if enterprises deploy Luna in the cloud and compare it with Chinese open-source models such as Zhipu AI's GLM 5.3, OpenAI's solution is "cheaper."

It should be noted that this statement is a public assertion by OpenAI management and does not disclose specific workloads, cloud provider quotes, or complete cost calculation methodologies; therefore, it cannot be directly equated to a universal conclusion across all enterprise scenarios.

However, this statement reveals a shift in the competitive dynamics of AI models: while open-source models typically do not charge model licensing fees, enterprises choosing self-deployment or deployment via cloud providers still bear costs for compute power, operations and maintenance, model fine-tuning, security management, and system integration. By contrast, the advantage of managed model services lies in more standardized usage and a cost structure that is easier to budget and manage.

As enterprise customers increasingly focus on AI return on investment, competition among model providers is no longer limited to the price per million tokens, but extends to overall deployment costs, engineering efficiency, and business outcomes.

Friar's statements also indicate that OpenAI is seeking to use low-cost models to compete for high-volume, high-frequency, and standardized enterprise application scenarios, while addressing complex reasoning and high-value tasks with higher-capability models. For enterprise customers, what really needs to be compared is not a single model's price quote, but the total cost of ownership comprised of model capabilities, call volumes, cloud resources, O&M investment, and business output.

Currently, OpenAI's enterprise revenue growth is outpacing its overall revenue growth, with its enterprise and consumer businesses approaching parity ahead of schedule, indicating that the commercial gravity of generative AI is shifting toward the enterprise segment. Cost competition between Luna and Chinese open-source models also signifies that the global large model market is further shifting from a capability race to a contest over deployment efficiency and return on investment.

This content was translated using AI and reviewed for clarity. It is for informational purposes only.