OpenAI Retires GPT-4o and Older ChatGPT Model Options

Announcement scope

OpenAI announced on January 29, 2026 that, starting February 13, 2026, ChatGPT would no longer support GPT-4o and certain older model selections. The notice says this applies across all ChatGPT user tiers. OpenAI also clarified that API support for existing models continues, indicating that lifecycle decisions for ChatGPT and API surfaces are being managed separately.

The model transition map is explicit in the announcement: GPT-4o is replaced by GPT-5, GPT-4.5 is replaced by GPT-5 thinking, and o4-mini is replaced by o5-mini. In practical terms, this is less about removing capacity and more about consolidating users onto a newer model lineup with a simpler selection framework.

Why this matters for teams

Model retirement policies are now a core operational mechanism for major AI platforms. Running many legacy options in parallel increases maintenance load across safety tuning, product consistency, latency routing, and cost control. Consolidation can improve platform-level quality and supportability, but it also shifts migration work to users.

For organizations with standardized prompt workflows, this transition should be treated as an operational change rather than a cosmetic UI change. Prompt templates, quality baselines, and acceptance criteria often drift when model families change. Even when capability improves, response style and tool-use behavior may differ enough to affect downstream automations.

Recommended migration checks

• Retest critical GPT-4o prompts against GPT-5 equivalents
• Benchmark GPT-5 thinking versus prior GPT-4.5 reasoning tasks
• Validate lightweight workflows moved from o4-mini to o5-mini
• Separate ChatGPT-facing and API-facing model governance policies

The broader signal is clear: AI product vendors are moving toward tighter model lifecycle cadence. Teams that build repeatable migration playbooks will absorb these transitions faster and with less production risk.