APEX 26.1 and APEXlang: the OAF-to-APEX business case just got stronger

Oracle APEX 26.1 became generally available on 14 May 2026, with the download files re-released on 25 May 2026 to address a bug found immediately after launch. If you are running Oracle E-Business Suite with a portfolio of OAF customizations, or a fleet of Forms applications that have been quietly holding the business together for a decade, this release is more relevant than the version number suggests. It changes the calculus of an OAF to APEX migration in a way that is worth understanding before your next planning cycle.

Oracle positions APEX 26.1 as enterprise low-code built for the AI era. In practice, that phrase covers a real shift in how APEX applications can be defined, reviewed and generated. The two most important pieces for anyone thinking about modernizing OAF are APEXlang and the governed generative development capabilities layered on top of it.

APEXlang is a new open, declarative, human-readable application specification language for APEX. It lets an APEX application be represented as files. Files that you can read, review, version, diff, scan and validate with the same tools you already use for every other codebase in your organization.

That sounds boring until you have tried to run a large APEX estate through a real engineering process. The historic answer to code review on APEX was the exported application file, which was accurate but painful to review and hostile to line by line diffing. APEXlang gives teams a supported way to make Git and version control workflows first class, which in turn makes serious peer review, change management and static analysis practical. It also gives large language models a supported, well defined representation of an APEX application to work with, rather than pattern matching against exported artefacts.

APEXlang is the foundation, and Oracle has built visible features on top of it in this release. The ones that matter most for enterprise teams:

• AI Agents, integrated into the APEX development experience for generation and refactoring against a governed application spec.
• Natural language queries in Interactive Reports, so end users can ask questions of a report in plain language against columns the developer has already exposed.
• Page creation from a natural language description, useful for scaffolding, still requiring a senior review before it lands in production.
• Data Reporter, a new self-service reporting capability where administrators curate trusted datasets and end users build their own reports on top of them.
• A new Iris theme style, a Metric Card component, Font APEX 2.5, and improved Content Security Policy support that make it easier to ship on-brand, secure applications by default.

APEX 26.1 requires Oracle REST Data Services 26.1.1 or later. That is a real prerequisite to plan for, not a footnote, because ORDS upgrades often need coordination with DBAs, load balancers and the identity stack. APEX itself continues to ship as a no-cost feature of the Oracle Database, supported under the database maintenance agreement, with the current release and the two prior releases supported at any given time. Two major APEX releases a year is roughly the cadence to plan against.

The historic objection to an OAF to APEX rebuild has always been the same one: rebuild effort and cost. OAF pages are often deceptively complex, with years of extensions, controllers, view objects and personalizations layered on top. The engineering cost of recreating that surface area, page by page, in any technology has been the reason so many organizations delayed and delayed.

APEXlang, combined with AI assisted development against a governed application spec, does move that number. When the target technology has a text-based, diffable representation, and when generation and refactoring can be done against that representation with proper review, the mechanical cost of producing an APEX page from a specification drops meaningfully. Combined with APEX 26.1's Metric Card, Iris theme and improved CSP support, the visual and security parity conversation with modern OAF pages also becomes shorter.

Here is where a senior view matters, because the marketing is confident and the reality is more nuanced. AI does not eliminate the hard part of OAF to APEX. The hard part has never been typing pages. The hard part is understanding what the OAF customization actually does inside EBS. Which base tables it reads and writes. Which standard controllers it extends. Which business rules were captured in a personalization ten years ago because it was faster than raising a service request. Which downstream reports and integrations depend on the exact behaviour of a specific button on a specific page.

None of that lives in the OAF page. It lives in the EBS data model, in customization history, and in the heads of the people who have been operating the system. A model can accelerate the rebuild, but it cannot discover the intent. Generated code still needs senior review, and every generated page still needs to be tested against real EBS data and real integrations before it goes near production. Teams that skip that step in the name of speed produce the next generation of customization debt, not a modernization.

For most organizations, especially those we work with across the GCC, the US, the UK, Europe and Australia, the right response to APEX 26.1 is not a large multi year programme. It is a fixed-price pilot on a single high pain OAF page. Choose the customization that hurts the most, the one that breaks at quarter close or that only one person understands, and rebuild it in APEX 26.1 with proper review, proper testing against your EBS data, and proper documentation. Prove the pattern on one page, on your own stack, with your own engineers watching.

From there, incremental modernization is a decision the business can size and sequence on real evidence, not on a slide. The EBS environment stays supported and stable in the background, the customization debt starts going down every quarter, and the modernized surface area grows on business merit rather than under pressure. APEX 26.1 does not remove the need for that discipline. It just makes the case for doing it now materially stronger than it was six months ago.