OpenAI says its newer models are substantially better at writing and migrating software built in older programming languages, including in legacy environments that many businesses still rely on. While the announcement is based on company statements rather than independent verification in the source, it is a notable development for organisations planning application modernisation, IT cost reduction, or internal automation projects.

What happened

According to the report, OpenAI said that newer models such as GPT-6.1 Sol show significantly improved performance in software tasks involving legacy code. That includes helping developers understand, rewrite, or migrate older applications into more modern stacks.

The claim matters because legacy migration has traditionally been expensive, slow, and dependent on scarce specialist skills. Many older systems still run core business processes in areas such as finance, logistics, manufacturing, and internal administration.

Why it matters for European businesses

For many European SMEs and mid-sized firms, older software is still embedded in daily operations. These systems may be stable, but they can also create problems:

• Higher maintenance costs when experienced developers for older languages are difficult to hire.
• Slower digital transformation when legacy applications are hard to connect with modern e-commerce, CRM, analytics, or automation tools.
• Security and compliance risks if unsupported systems are difficult to patch or document.
• Operational dependency on a small number of internal experts or external contractors.

If AI models become more capable at analysing and translating older codebases, businesses may be able to accelerate some stages of modernisation. That could include code documentation, refactoring, test generation, API integration planning, and migration support.

However, this does not mean AI can automatically replace a structured migration programme. Legacy systems often include undocumented business logic, custom workflows, and regulatory requirements that need human review.

Who may be affected

• SMEs running older internal systems for accounting, inventory, production, or customer management.
• E-commerce companies that depend on older back-office tools and want better integration with modern storefronts, payment systems, or marketing platforms.
• IT teams and digital agencies supporting clients with application upgrades, technical debt reduction, or cloud migration.
• Founders and business owners evaluating whether to rebuild, replace, or gradually modernise ageing software.
• Regulated businesses that must document systems carefully and cannot accept untested AI-generated code in critical workflows.

What companies should consider

Businesses interested in using AI for legacy migration should treat the development as promising, but not as proof that complex upgrades can be safely automated end to end.

• Start with low-risk pilot projects such as documentation, code explanation, dependency mapping, or test creation.
• Validate all outputs through experienced developers, especially where business-critical logic is involved.
• Review security and privacy controls before sharing proprietary source code with external AI services.
• Assess integration value by identifying which legacy components create the biggest cost, security, or growth constraints.
• Keep governance in place with version control, audit trails, testing, and clear approval processes for any AI-assisted code changes.
• Check contractual and compliance requirements if applications process personal data, financial records, or sector-specific regulated information.

For European companies, the practical opportunity is not simply faster coding. It is the possibility of reducing technical debt and making older systems easier to secure, integrate, and automate. But the business case will depend on whether these model improvements translate into reliable results in real migration projects.