SAN FRANCISCO — January 2026 — OpenAI has announced plans to retire GPT-4o, GPT‑4o, GPT‑4.1, GPT‑4.1 mini, and OpenAI o4-mini models on February 13, 2026, marking a transition toward newer, more capable systems across its API and product ecosystem.

The move is part of OpenAI’s ongoing effort to streamline its model lineup, reduce fragmentation, and focus development on models that deliver stronger performance, lower latency, and improved safety characteristics. OpenAI said maintaining a smaller set of more capable models allows it to iterate faster and provide clearer guidance to developers and customers.

GPT-4o, which introduced multimodal capabilities and faster responses compared to earlier GPT-4 models, will be phased out alongside other legacy models that are no longer aligned with OpenAI’s long-term roadmap. The company emphasized that newer models already offer equal or better performance across most use cases.

Developers are being encouraged to migrate to OpenAI’s latest generation models, which provide improved reasoning, efficiency, and tooling support. OpenAI said it is offering advance notice and documentation to help teams transition smoothly, minimizing disruption to production systems.

The retirement plan reflects a broader pattern in AI platform development, where rapid advances make it impractical to support many overlapping model generations. OpenAI said focusing resources on fewer models improves reliability, security updates, and overall user experience.

OpenAI did not position the change as a reduction in capability, but as an evolution of the platform. As new models take on a wider range of tasks with better performance and cost efficiency, older systems are being sunset to keep the ecosystem simpler and more sustainable.

The company said timelines and specific retirement dates will be communicated clearly to affected users, giving developers time to test and deploy replacements before models are fully deprecated.

With the decision, OpenAI is signaling that AI platforms are entering a more mature phase, where model lifecycle management and long-term maintainability are becoming as important as raw capability gains.