GPT-4, the AI model that powered ChatGPT and sparked widespread adoption of advanced AI assistants, is no longer available in OpenAI's consumer product as of 2026. The model has been fully retired from ChatGPT's interface, though it remains accessible to developers through OpenAI's API. This represents a significant milestone in how quickly AI technology evolves and how companies manage the lifecycle of their products.

The retirement didn't happen all at once. OpenAI phased out GPT-4 and its variants in stages between February and June 2026. On February 13, 2026, the company removed GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini from ChatGPT's model picker for most users, citing that usage had shifted heavily toward newer GPT-5-series models. Business, Enterprise, and Education customers retained access to GPT-4o within Custom GPTs until April 3, 2026, after which it was fully retired across all subscription tiers. Finally, GPT-4.5, the last remaining GPT-4-family model still available in ChatGPT, was retired in late June 2026.

What Was GPT-4 and Why Did It Matter?

GPT-4 was a large language model developed by OpenAI and first released in March 2023. It represented a substantial capability jump over the earlier GPT-3.5 model that powered ChatGPT at its initial public launch. The model became significant because it offered better reasoning, more reliable instruction-following, and stronger performance across professional and creative tasks. When it launched, GPT-4 was widely discussed enough that "ChatGPT 4" became shorthand for "the advanced version of ChatGPT" in general public usage.

The model's documented capabilities included several features that made it stand out for its time. These capabilities shaped how millions of people used AI for everyday tasks and professional work.

• Advanced language understanding: More nuanced comprehension of complex or ambiguous requests compared to earlier models
• Improved reasoning: Better performance on tasks requiring multi-step logical thinking and problem-solving
• Longer context handling: Ability to work with longer documents and conversations than earlier models allowed
• Multimodal capability: Image understanding in specific products and configurations where enabled
• Stronger coding assistance: More reliable code generation and debugging support across multiple programming languages
• Better instruction-following: More consistent adherence to specific formatting or task requirements

How Did People Actually Use GPT-4?

GPT-4 was widely used across multiple domains, from education to business to creative work. The same broad categories of tasks people use current AI models for today were powered by GPT-4 as the engine behind them at the time. Common use cases included writing assistance, coding help, research support, business communication, and educational explanation.

Professionals and students relied on GPT-4 for specific, practical tasks. Someone might ask it to explain technical SEO to a beginner using a simple example, or request that it review a Python function and identify potential bugs. Content creators used it to generate 30-day content plans for e-commerce brands. These same prompt patterns work with current ChatGPT models today, because the underlying skill of writing clear, specific prompts hasn't changed even as the model powering the responses has evolved.

What's Available in ChatGPT Now?

If you open ChatGPT today, you cannot select any GPT-4 variant in the model picker. Old conversations that used a GPT-4 model now default to the current equivalent GPT-5-series model. Custom GPTs built on retired models automatically moved to current model equivalents. For anyone using ChatGPT directly through the web interface or mobile app, the current default experience now runs on GPT-5-series models, which OpenAI positions as offering stronger reasoning capabilities and more customization options.

This shift reflects a broader pattern in AI product development. ChatGPT is the chat application; GPT-4, GPT-5, and other model names refer to the underlying engines that power it at different points. Saying "I use ChatGPT" doesn't tell you which model is actually generating your responses, since that has changed multiple times since GPT-4's original release. The ChatGPT web experience doesn't guarantee access to every model OpenAI has ever released. Models get added, updated, and retired as the product evolves, and what's available to you specifically can also depend on your subscription plan, whether you're on the Free, Go, Plus, Pro, Business, or Enterprise tier.

How to Understand ChatGPT's Current Model Lineup

• Check your subscription tier: Your access to specific models depends on whether you're using the free version or a paid plan like Plus, Pro, Business, or Enterprise
• Look at the model picker: When you start a new conversation in ChatGPT, you can see which models are available to you in the interface dropdown menu
• Understand that models change: OpenAI regularly updates which models are available, retires older ones, and introduces new versions, so your options may differ from what a friend sees
• Know that old conversations migrate: If you have previous conversations that used GPT-4, they automatically switched to the current equivalent model when GPT-4 was retired
• Access GPT-4 through the API if needed: Developers can still access GPT-4-family models through OpenAI's developer API, which was not affected by these ChatGPT-specific retirements

The retirement of GPT-4 from ChatGPT marks a natural progression in AI product evolution. As newer models with improved capabilities become available, companies phase out older versions to streamline their product offerings and encourage users to adopt more advanced tools. For most ChatGPT users, this transition happened seamlessly in the background, with conversations and custom tools automatically adapting to the new models without requiring any action on their part.