When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works.

OpenAI improves STEM answers in updated GPT-4o model and increases emoji use

OpenAI logo
OpenAI

OpenAI has announced that its GPT-4o model has been updated with more up-to-date information, better abilities to analyze images and improvements in answers related to STEM fields. The company said that the model now uses emojis more often to make flare-up responses a bit more.

While OpenAI has now given ChatGPT users web access, the underlying models themselves still have a cut-off date, which can affect their responses. In this latest update, the cut-off date has been moved up from November 2023 to June 2024. This will improve its social and cultural understanding somewhat. It can also enhance its answers from web results.

“GPT-4o is now better at understanding and answering questions about visual inputs, with improvements on multimodal benchmarks like MMMU and MathVista,” OpenAI said, outlining another improvement with this updated model.

“The updated model is more adept at interpreting spatial relationships in image uploads, as well as analyzing complex diagrams, understanding charts and graphs, and connecting visual input with written content. Responses to image uploads will contain richer insights and more accurate guidance in areas like spatial planning and design layouts, as well as visually driven mathematical or technical problem-solving.”

For anyone interested in STEM-related topics, this new model improves its answers in this department, too, thanks to gains on academic evals like GPQA and MATH. It also does better on MMLU, reflecting its ability to tackle more complex problems across domains. So, if you use ChatGPT for math, science, or coding help, then expect better results.

Finally, to add a bit of flare to its responses, GPT-4o will now use more strategically placed emojis to make answers a bit more colorful and engaging. All these changes seem to be only for GPT-4o right now rather than the 4o-mini model. Hopefully, that model will get updated, too, in the near future, as it's what most people on the free tier are using.

Report a problem with article
The Perplexity logo
Next Article

Perplexity could be about to drop a big reasoning update for its new assistant

DeepSeek
Previous Article

DeepSeek sees partial restoration of services after massive DDoS attack

Join the conversation!

Login or Sign Up to read and post a comment.

3 Comments - Add comment