We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
OpenAI Blog·Jan 29·2 min read
Retiring GPT-4o, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini in ChatGPT
TL;DR
OpenAI is discontinuing older large language models (LLMs—AI systems trained on vast text data) from ChatGPT to streamline its product lineup. GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini will be retired on February 13, 2026, alongside previously announced GPT-5 variants.
✦ Why It Matters
Engineers using ChatGPT must plan model migrations; API users face no immediate action required.
Key Takeaways
- Update ChatGPT workflows to use GPT-5 variants before February 13, 2026 retirement date.
- Verify API integrations remain unaffected; no immediate action required for backend systems.
- Plan migration strategy for production applications still using retired models.
- Monitor OpenAI announcements for future model deprecation timelines and replacements.
Full Summary
OpenAI announced the retirement of multiple language model versions from its ChatGPT consumer product effective February 13, 2026. The discontinued models include GPT-4o (a multimodal model supporting text and images), GPT-4.1, GPT-4.1 mini (a smaller, faster variant), and o4-mini, alongside previously announced GPT-5 variants (Instant, Thinking, and Pro versions).
This represents standard product lifecycle management where older model versions are removed as newer alternatives mature. Importantly, the retirement applies only to ChatGPT's user-facing interface; the API (application programming interface—the programmatic access layer for developers) remains unaffected, meaning engineers can continue integrating these models into applications.
This distinction allows OpenAI to streamline consumer offerings while maintaining backward compatibility for production systems. The move suggests GPT-5 variants are now sufficiently stable and performant to replace their predecessors.