OpenAI Unveils Advanced GPT-4.1 AI Models
OpenAI has launched its latest series of artificial intelligence models, the GPT-4.1 family, which includes three variants: GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano. Announced on Monday, these models promise significant enhancements over their predecessors, GPT-4o and GPT-4o mini, particularly in coding capabilities and instruction adherence. However, access to these models is limited to developers through an application programming interface (API), as they will not be available via ChatGPT.
OpenAI Releases Coding-Focused GPT-4.1 AI Models
OpenAI’s new GPT-4.1 series is designed specifically for coding tasks and complex instruction-following scenarios. The company has highlighted several improvements over the previous GPT-4o models, including enhanced performance metrics, a larger context window, and more competitive pricing. Notably, the knowledge cutoff for these models has been updated to June 2024, ensuring they are equipped with the latest information.
In terms of performance, the GPT-4.1 model achieved an impressive score of 54.6 percent on the SWE-bench Verified benchmark, surpassing both GPT-4o and the newly released GPT-4.5. This positions GPT-4.1 as OpenAI’s top model for coding-related applications. Additionally, it scored 38.3 percent on the MultiChallenge benchmark for instruction adherence and 72 percent on the Video-MME benchmark for understanding multimodal long contexts. Each model in the GPT-4.1 series supports a context window of up to one million tokens, enhancing their usability for developers.
Performance and Pricing Structure
While the GPT-4.1 model stands out as the flagship offering, the GPT-4.1 mini also shows significant improvements over the GPT-4o. OpenAI claims that the mini variant reduces latency by 50 percent and is 83 percent cheaper than its predecessor. This makes it an attractive option for developers looking for efficiency without compromising performance.
The GPT-4.1 nano model is tailored for low-latency tasks, boasting a score of 80.1 percent on the MMLU benchmark, 50.3 percent on GPQA, and 9.8 percent on the Aider polyglot coding benchmark. OpenAI asserts that the nano model outperforms the GPT-4o mini, making it suitable for a variety of applications. The enhanced instruction-following capabilities and extended context window further position these models as viable options for powering advanced AI agents.
In terms of pricing, the GPT-4.1 model is available at $2 (approximately Rs. 171) per million input tokens and $8 (approximately Rs. 685) per million output tokens. The mini variant is priced at $0.40 (approximately Rs. 34) per million input tokens and $1.60 (approximately Rs. 137) per million output tokens, while the nano model is the most economical at $0.10 (approximately Rs. 8.5) per million input tokens and $0.40 (approximately Rs. 34) per million output tokens.
Availability and Future Developments
It is important to note that the GPT-4.1 series will only be accessible to developers through the API, limiting its availability to a select group. However, OpenAI has indicated that the advancements seen in the GPT-4.1 models will eventually be integrated into the latest version of GPT-4o, which remains available through ChatGPT. This transition aims to ensure that users of ChatGPT can benefit from the latest improvements in AI technology.
Observer Voice is the one stop site for National, International news, Sports, Editor’s Choice, Art/culture contents, Quotes and much more. We also cover historical contents. Historical contents includes World History, Indian History, and what happened today. The website also covers Entertainment across the India and World.