OpenAI Updates GPT-4o: Rankings Jump, New Advanced Capabilities | OpenAI ChatGPT | OpenAI italiano | OpenAI Login | Turtles AI

OpenAI recently updated GPT-4o, bringing the model to #2 in Chatbot Arena and significantly improving its performance in math, coding, and complex prompts.

Key Points:

• Rankings Improvements: GPT-4o rose from #14 to #1 in math and from #7 to #1 in difficult prompts.
• Coding Performance: The model rose from #5 to #1 in code generation.
• Availability: The update is available immediately to paid ChatGPT users; free users will receive it in the coming weeks.
• API Access: Developers can access the new model via the “chatgpt-4o-latest” API.

OpenAI recently introduced a significant update to its GPT-4o language model, resulting in a significant improvement in benchmarking rankings. The updated model reportedly reached #2 in Chatbot Arena, even surpassing GPT-4.5, which was launched last month. In particular, GPT-4o showed impressive gains in key categories: in math, it jumped from 14th to 1st; in complex prompts, from 7th to 1st; and in coding, from 5th to 1st.

OpenAI says the updated model excels at following detailed instructions, especially when prompts contain multiple requests. It also improved its ability to tackle complex technical and coding problems, generating simpler and cleaner front-end code, more accurately identifying necessary changes to existing code, and producing output that compiles and runs successfully.

Early testers report that the model appears to better understand the implicit intent of their prompts, especially in creative and collaborative tasks. Responses are more concise and clear, with fewer complex markdown formatting and emojis. Sam Altman, OpenAI’s CEO, described the improvements as substantial and teased more to come.

The GPT-4o update is available now for all paid ChatGPT users, and free users will receive it in the coming weeks. For developers, the new model is accessible via the “chatgpt-4o-latest” API. OpenAI also plans to extend these improvements to older models in the API in the near future.

These developments highlight OpenAI’s ongoing commitment to improving the capabilities of its language models, giving users increasingly powerful tools to tackle a wide range of complex tasks.