IBM and OpenAI have expanded their collaboration with a strategic partnership aimed at bringing the company’s artificial intelligence models and tools into business processes that range from software development to finance, human resources, and cybersecurity. Announced on August 13, the agreement will incorporate GPT-5.6, Codex, and ChatGPT Work into IBM Consulting Advantage and will be supported by a new specialized IBM practice with thousands of consultants and engineers trained through the OpenAI Partner Network.
The key points of the IBM and OpenAI partnership in 20 seconds
- IBM will integrate GPT-5.6, Codex, and ChatGPT Work into IBM Consulting Advantage.
- They will establish an OpenAI Practice with thousands of specialized consultants and engineers.
- The partnership targets banking, government, telecommunications, and commerce, as well as internal processes.
- A significant focus will be on modernizing legacy applications and systems.
- The collaboration extends the work both companies have been doing in cybersecurity since June.
This move is of interest beyond just adding another business partner to OpenAI. IBM is positioning its models alongside proprietary technologies and those from other providers within a platform it already uses to develop AI projects. Consulting Advantage currently equips nearly 150,000 IBM consultants with assistants, agents, and sector-specific applications.
The company is not suggesting that all companies replace their current infrastructure with a single platform. Its AI services strategy has maintained a hybrid, multi-model approach supported by different clouds and both open and proprietary models. For example, in January 2026, IBM launched Enterprise Advantage with compatibility for AWS, Google Cloud, Microsoft Azure, IBM watsonx, and various model families.
Adding OpenAI today adds another significant component to this ecosystem.
GPT-5.6 and Codex arrive in legacy enterprise systems
One of the prioritized goals is likely also one of the least glamorous challenges in enterprise AI adoption: connecting advanced models with applications, processes, and data that have been operational for years or decades.
IBM and OpenAI intend to use Consulting Advantage to analyze procedures and workflows to identify where AI integration can be implemented. The initial focus areas include finance, procurement, customer service, and human resources.
This sets the project apart from simply deploying a corporate chatbot.
A company can have access to advanced models while still having much of its activity distributed across ERP systems, internally developed applications, databases, SaaS tools, and manual processes.
The challenge then shifts from simply gaining access to the model.
Decisions must be made about what information the model can query, what actions it can perform, who authorizes those actions, how they are audited, and what happens if the model makes a mistake.
Andy Baldwin, IBM Senior Vice President of Global Consulting, summarized this approach by stating that the business challenge is no longer just about accessing AI technology, but about integrating it securely and at scale within complex environments and processes.
To work in this domain, IBM will also create teams of specialists deployed directly with clients. These teams will be formed through the OpenAI Partner Network and will aim to adapt systems to AI-ready workflows.
The company will also launch a dedicated OpenAI Practice, expecting thousands of consultants and engineers to earn advanced certifications through the OpenAI Partner Program. Additionally, IBM will join the Elite level of this network.
Codex makes legacy software modernization another element of the deal
Another aspect is application development.
IBM will combine Codex and ChatGPT Work with its consulting services to modernize legacy software and develop new digital products. The company has not disclosed specific time or cost savings goals for this collaboration, so the real-world results remain to be seen.
However, IBM’s starting point already has a track record.
Consulting Advantage launched in 2024 as an internal platform providing its consultants with specialized assistants. IBM reported productivity improvements of up to 50% in a pilot related to application design, development, and testing—though these results were from a specific pilot and not automatically applicable to all projects.
Since then, the platform has evolved into a much broader model.
In January 2026, IBM confirmed that Consulting Advantage had participated in over 150 projects with clients. Subsequently, Enterprise Advantage extended part of this approach to organizations seeking to build their own internal AI platforms.
The integration of Codex fits particularly well within this evolution, as AI-assisted programming is shifting from completing code snippets to delegating more extensive tasks to agents.
A modernization project might include analyzing existing code, generating documentation, updating dependencies, creating tests, or migrating components. However, the output still requires validation, especially when it impacts critical enterprise applications.
The partnership aims to embed these capabilities within a consulting process rather than simply providing direct agent access.
Cybersecurity is the third front for IBM and OpenAI
This agreement is not starting from scratch.
In June, IBM joined the OpenAI Daybreak Cyber Partner Program, an initiative aimed at using advanced models for cybersecurity defense. As part of that announcement, IBM introduced an application security service that employs OpenAI models to identify and validate vulnerabilities.
Two months later, the two companies plan to expand this collaboration.
The new partnership will combine OpenAI models with IBM Autonomous Security, IBM’s multi-agent service designed to coordinate security decisions, intelligence, and response.
The stated goal includes addressing both traditional threats and risks associated with AI adoption itself: application vulnerabilities, governance issues, and operational risks related to models.
This is especially relevant as agents move beyond merely answering questions.
An assistant that only drafts a document has a limited scope. An agent connected to repositories, development tools, financial systems, or enterprise applications can execute actions, which necessitates reviewing permissions, identities, logs, and controls.
Therefore, security becomes part of the agent’s architecture, not just its underlying infrastructure.
IBM is deliberately building a multi-model strategy
The partnership also reveals how IBM is positioning itself in the AI market.
OpenAI will not be the only external provider integrated into their services.
On June 4, IBM announced a new global practice with Google Cloud that also includes Gemini Enterprise within Consulting Advantage. This initiative involves thousands of specialized consultants and industry-specific agents.
IBM also maintains watsonx and its own Granite models.
In 2024, IBM made Granite 3.0 the default model for Consulting Advantage, while the platform continued to support integrations with partner technologies.
Thus, the addition of GPT-5.6 should not be seen as replacing Granite or abandoning IBM’s original strategy.
Instead, it appears to be about creating a layer of consulting, integration, and governance capable of working with multiple model providers.
This approach makes particular sense for large organizations.
A bank might use one model for software development, another for specific internal tasks, and proprietary infrastructure for highly sensitive data. Decisions may depend on cost, performance, data location, security, or regulation.
The challenge arises when all these pieces need to function within unified processes.
Exactly where IBM aims to position its business.
OpenAI also needs deeper penetration into large enterprises
From OpenAI’s perspective, the partnership has a second layer of meaning.
Developing an advanced model does not guarantee it will be integrated into the core systems of a bank, government, or telecom operator.
These organizations require integration, migrations, access controls, governance, compliance, and teams capable of maintaining deployed solutions.
IBM provides exactly that presence.
With operations in more than 175 countries, IBM has longstanding relationships with governments, banks, utilities, and organizations managing critical infrastructure.
Meanwhile, OpenAI is strengthening its enterprise deployment structure in 2026. In May, it launched OpenAI Deployment Company, a dedicated organization to help companies bring advanced AI systems into production.
The collaboration with IBM aligns with this direction but offers something different: a network of consultants already working within many of the organizations OpenAI aims to reach.
The outcome also reflects where enterprise AI competition is heading.
Having access to the most advanced model remains important, but increasingly it explains less on its own.
The next frontier is connecting these models with real applications, corporate data, legacy systems, security tools, and processes where a wrong decision can have financial or operational consequences.
IBM and OpenAI have chosen to walk part of that path together.
Now, they will need to demonstrate that integrating GPT-5.6, Codex, and ChatGPT Work into a platform used by thousands of consultants yields more than just another impressive AI demo.
The real test will be in production.
Frequently Asked Questions
What has IBM announced with OpenAI?
IBM and OpenAI have formed a commercial and technical alliance to integrate OpenAI models and products into enterprise projects. IBM will incorporate GPT-5.6, Codex, and ChatGPT Work into IBM Consulting Advantage and establish a specialized OpenAI practice.
Will IBM replace watsonx or Granite with OpenAI models?
No such substitution has been announced. IBM maintains a multi-model strategy that includes its Granite models and technologies from other providers, along with partnerships like Google Cloud.
How will IBM use Codex?
IBM plans to use Codex for application modernization projects and software development, combined with the technical and sector expertise of its consulting teams.
Did IBM and OpenAI collaborate before?
Yes. In June 2026, IBM joined the OpenAI Daybreak Cyber Partner Program to apply advanced models to cybersecurity tasks. The new partnership expands this relationship to more enterprise processes.
