GPT 4.1 Update: New Developer Model Now Integrated in ChatGPT!

GPT-4.1 Now Available in ChatGPT

In a recent announcement on X this Wednesday, May 14, OpenAI revealed that the GPT-4.1 model has been made available in ChatGPT. Initially unveiled in mid-April, it was exclusively accessible through the API in three versions: GPT-4.1 (full and powerful, suited for complex cases), GPT-4.1 mini (lightweight and cost-effective, ideal for high-performance AI tools), and GPT-4.1 nano (extremely fast, designed for simple and frequent tasks).

GPT-4.1 is a language model tailored for developers, designed to achieve “significant improvements in coding and instruction adherence,” according to the company. Specifically, the latest iteration offers substantial advancements in three key areas:

• Code: GPT-4.1 efficiently corrects code, successfully handling 54.6% of cases in the internally designed SWE-bench Verified benchmark, significantly outperforming previous models.
• Instruction Understanding: GPT-4.1 shows great improvement in following complex instructions, with a 10.5-point increase on the MultiChallenge test compared to earlier versions.
• Long-Term Multimodal Understanding: GPT-4.1 excels in comprehending long videos without subtitles, achieving a record-breaking score on the Video-MME benchmark (72% success rate).

Additionally, OpenAI has released the safety report for GPT-4.1, addressing the absence of such a report during the model’s initial API launch, which had drawn significant criticism from the research community.

How to Access the GPT-4.1 Model in ChatGPT

Two versions have been incorporated into the conversational agent for users on the Plus, Pro, and Team paid plans: GPT-4.1 and GPT-4.1 mini. In its release notes, OpenAI states that the latter will replace GPT-4o mini.

GPT-4.1 mini is a small, fast, efficient, and high-performing model, offering significant improvements over GPT-4o mini in terms of instruction adherence, coding, and overall intelligence. Effective immediately, GPT-4.1 mini will replace GPT-4o mini for paying users. It will also serve as a backup model for free users once they have exhausted their usage limits of GPT-4o, OpenAI explains.

To access GPT-4.1 and GPT-4.1 mini, go to the model selector in the upper-left corner of the interface, and hover over the More Models section (refer to the featured image).