OpenAI published the framework on July 17, 2026 under the concept “Useful Intelligence per Dollar.” It asks whether AI completes work that matters, what each successful task costs, whether users can depend on the result and whether each dollar produces more value as usage grows.
The central proposal is to measure the full cost of a successful outcome. That includes model usage, employee time, human review, retries and rework. OpenAI argues that a model with a higher token price can still be cheaper if it completes a task correctly in fewer attempts and requires less intervention.
Three GPT-5.6 tiers connect model choice with workflow economics
OpenAI used its GPT-5.6 family to illustrate the framework. The company offers Sol as its flagship tier, Terra as a balance between performance and cost, and Luna as its fastest and most affordable tier. It recommends choosing among them according to the economics and quality requirements of the full task.
OpenAI reported that GPT-5.6 Sol, using maximum reasoning, consumed 54 percent fewer output tokens than another leading model on the Artificial Analysis Coding Agent Index. It also reported a 72.7 percent result on DeepSWE v1.1, compared with 69.9 percent for Claude Fable 5, at an estimated API cost that was 36.2 percent lower. These are company-presented benchmark claims and should be evaluated in the context of the tested workloads.
Four operating measures move AI buying toward outcomes
The framework starts with a defined workflow and a clear definition of completion. Examples include a customer issue resolved, a code change that passes tests or a contract reviewed accurately and on time. OpenAI then suggests classifying outputs as ready to use, needing correction or needing escalation.
The company also says organizations should define what data an AI system can access, which systems it can change and when human approval is required. That connects financial measurement with governance because low apparent task costs can be misleading when failures create security, compliance or customer risks.
Analysis: outcome pricing could reshape SaaS product design
Editorial analysis: For SaaS founders, the framework supports pricing and product decisions based on completed work rather than access alone. The opportunity is to prove value through resolution rates, successful automations, accepted code changes or verified decisions. Investors can also compare growth with the cost and reliability of the work delivered, rather than treating AI usage as a sufficient indicator of product-market fit.
More information
Source and more information: OpenAI