OpenAI Urges Enterprises to Use Its Scorecard to Measure Worth of AI

Enterprises concerned about the potential cost of buying and implementing AI should rethink how they evaluate its worth, according to OpenAI.To assist them in doing so, the frontier AI lab has published a “scorecard for the AI age” designed to maximize the value of their AI investment.“For years, the market measured the success of software through adoption: seats purchased, users active, licenses renewed,” CFO Sarah Friar said in a blog post. “Understanding the value of AI demands a more powerful measure: work accomplished.”This change in mindset looks beyond cheap tokens and instead places more priority on examining what a model has delivered and how it has done so. According to Friar, models with affordable tokens do not always equate to value, sometimes taking longer than more expensive models to fulfill a task.That’s why OpenAI is pushing its metric, “Useful Intelligence Per Dollar,” in which “what matters is the full cost of producing a successful outcome, measured against the value that outcome creates,” the blog post said.Related:Alibaba Qwen 3.8 Max Shows China Closing in on U.S. ModelsThe scorecard is based on a number of principles.First, OpenAI wants enterprises to measure how much useful work they are doing. This requires businesses to define what “done” actually constitutes within their operations and compare this outcome with how things were achieved before AI.Second, Friar argued that enterprises need to consider the cost of successful outcomes, taking into account employee time, human review, retries, and rework. By introducing this methodology, a business would gain insight into the level of model required. ChatGPT-5.6, for example, has three tiers -- Sol (the most comprehensive), Terra (mid-level) and Luna (the most affordable).A third question enterprises should be considering is dependability, according to OpenAI. Friar said that accurate and consistent results require less input from human staff, saving a business money in the long run. This, she argued, has “direct economic value.”Finally, OpenAI wants enterprises to consider whether investment buys more work as usage grows -- or, in other words, whether the economics improve at scale.By tracking outcomes, total cost and cost per outcome over time, businesses can see whether their AI investment is gaining value.The value of introducing AI to enterprises has faced sharper scrutiny than ever of late, with a PWC study in January revealing that only 12% of CEOs think it has delivered cost and revenue benefits.About the AuthorContributing WriterGraham Hope has worked in automotive journalism in the U.K. for 26 years, including spells as editor of leading consumer news website and weekly Auto Express and respected buying guide CarBuyer.