Launch of GPT‑4.1, GPT‑4.1 mini, and GPT‑4.1 nano via API
- All models outperform GPT‑4o and GPT‑4.5 in coding, instruction following, long context
- Up to 1 million token context window
- Superior instruction following
- Strong long-context & multi-hop reasoning
- GPT‑4.1 mini: matches/beats GPT‑4o, 2× faster, 83% cheaper
- GPT‑4.1 nano: fastest, cheapest, 1M context, strong at autocompletion/classification
- Output token limits increased (32,768 tokens for GPT‑4.1)
- Major improvements in frontend coding, format-following, comprehension
- Industry partners report large real-world gains
- Enhanced image and video understanding in all models
- GPT‑4.5 Preview API deprecated by July 2025
- Focus on reliability and agentic use cases (AI agents, tools)