OpenAI scales Codex to enterprise: Codex Labs program and 4 million weekly active users

OpenAI has officially announced the next phase of Codex expansion — its AI agent for software development — toward large corporate clients. The company unveiled the Codex Labs program, strategic partnerships with leading consulting firms, and announced that the tool has passed 4 million weekly active users.

What happened?

OpenAI introduced Codex Labs, a partner program connecting OpenAI with global consulting firms Accenture, Deloitte, and KPMG. The program’s goal is to accelerate enterprise adoption of Codex through certified experts, joint deployment packages, and industry-specific implementations.

Alongside the partner program, OpenAI announced that Codex has surpassed 4 million weekly active users — a significant jump in adoption since the tool expanded beyond the developer terminal. Codex is now used in IDEs, in the cloud, on mobile devices, and through GitHub integrations.

For enterprise clients, OpenAI is introducing new deployment packages with consumption-based pricing (billing based on actual usage), representing a shift away from the classic per-seat model. Consultants who complete certification receive the Codex specialist designation and access to advanced tools for client code migration.

Why does this matter?

This move marks a turning point in OpenAI’s commercial strategy — the company is no longer pushing only ChatGPT Enterprise as its primary enterprise product, but is simultaneously building a dedicated ecosystem around agentic software development. The partnership with the “big three” consulting firms gives Codex a distribution channel to thousands of large clients who already have contracts with Accenture, Deloitte, or KPMG.

The figure of 4 million weekly active users places Codex in direct competition with GitHub Copilot — which has until now been the dominant AI assistant for software engineers. The difference is that Codex goes beyond autocomplete suggestions and offers a full agentic workflow: the agent can independently take on a task, search the codebase, run tests, and open pull requests.

The consumption-based billing model is an additional signal — OpenAI expects that enterprise users will generate far more agentic work per person than a flat fee would cover, and conversely, that it will be cheaper than a fixed subscription for some teams.

What do the enterprise deployment packages look like?

Enterprise packages within the Codex Labs program include several standardized components. The first layer is a migration package — a certified consultant analyzes the existing codebase, identifies candidates for agentic automation (framework version migrations, legacy code modernization, dependency updates), and creates an implementation plan.

The second layer is a governance layer with tools for auditing agent actions, role-based access control to code repositories, and integration with the client’s existing CI/CD systems. This addresses the primary concern of enterprise security teams — that an agent must not have unrestricted access to production systems without oversight.

The third layer is training and enablement — certified consultants run internal workshops, help establish internal guidelines for Codex usage, and measure return on investment (ROI) through agreed metrics such as reduced time to merge request or the number of automated migrations.

What’s next?

The first waves of Codex Labs implementations are already underway at selected clients of Accenture, Deloitte, and KPMG, with a focus on financial services, healthcare, and manufacturing. OpenAI is announcing an expansion of the certification program to regional system integrators throughout 2026.

For the regional market, this means Codex will soon become available through local partners of the major consulting firms, opening doors even for mid-sized companies that would otherwise lack the resources for independent implementation of agentic tools. At the same time, smaller agencies and freelance consultants must stay current, as Codex specialist certification will become a significant market differentiator.

The key open question remains how enterprise clients will handle code security, auditing of agent actions, and compliance with regulations such as the EU AI Act. OpenAI has announced additional governance and agent monitoring tools as part of the Codex Labs package, but implementation details for the European market have not yet been publicly disclosed. OpenAI is expected to publish additional technical documents and reference architectures for regulated industries in the coming months.