OpenAI has launched GPT-4.1, the next evolution of its multimodal AI systems, following last year’s release of GPT-4o. In a recent livestream, the company shared that GPT-4.1 outperforms GPT-4o across nearly every category, with major improvements in areas like coding and instruction-following. One of the standout features is the expanded context window — GPT-4.1 can now handle up to 1 million tokens, a significant jump from GPT-4o’s 128,000-token limit. According to OpenAI, the model is more reliable at identifying relevant text and filtering out distractions, even with very long inputs.

Three New Versions, Lower Cost, and More Flexibility

GPT-4.1 is available now for developers, along with two smaller versions: GPT-4.1 Mini and GPT-4.1 Nano. The Mini version is designed to be budget-friendly for experimentation, while Nano is OpenAI’s lightest and fastest model so far. All three models support the full 1 million-token context length. GPT-4.1 is also 26% cheaper than GPT-4o, which matters more now as new efficient models like DeepSeek gain attention.

Shifting Release Plans and What’s Next

The debut of GPT-4.1 comes as OpenAI begins retiring older models. GPT-4 will be removed from ChatGPT on April 30th, and the GPT-4.5 preview will be phased out from the API on July 14th. GPT-4.1 is seen as the natural replacement, offering better performance at lower cost and latency. Meanwhile, GPT-4o, the current default in ChatGPT, was recently updated with image-generation features — so popular that OpenAI had to temporarily limit access to protect server capacity. CEO Sam Altman also confirmed that GPT-5 has been delayed by a few months, due to challenges integrating new features smoothly. OpenAI is expected to launch more models soon, including o3 and o4 mini reasoning models.