Early this morning, ChatGPT suddenly updated - the data analysis function based on Python is now available in o1 and o3-mini.

OpenAI introduced that users can now call Python through two models to complete tasks such as data analysis, visualization, and scenario-based simulation.

Quantum Bit's test found that although the interface has not changed significantly, Python has indeed been "on duty" quietly in o1.

Some netizens commented that this is equivalent to the third version of the ADA (Advanced Data Analysis) function:

Others evaluated that o1 has evolved from an assistant to an all-round data analysis Copilot.

o1 Challenges Python Data Analysis

In the first round of testing, we organized more than 400 flight records of a certain aircraft over nearly a year and handed them over to the model for data analysis.

The model successfully analyzed all four questions.

Of course, 4o also answered these questions correctly, but the output format was not very neat.

Claude (3.7 Sonnet), on the other hand, made some blunders, initially claiming it could not open the uploaded file (possibly due to the Chinese characters in the file name), and then repeatedly retrying and eventually making up its own answers.

After removing the Chinese characters from the file name, Claude successfully read the file and gave the correct conclusion.

After testing the performance on Q&A tasks, let's take a look at the visualization capability.

o1 (left) and 4o (right) drew the line graphs of altitude and speed changes over time after calling Python.

Claude's operation is more advanced, directly making an interactive web page, but there are problems with the time zone conversion and the lack of dashed lines for data points with longer time intervals.

One More Thing

In addition to the function updates of o1 and o3-mini, OpenAI also announced that the previously Mac client member-exclusive Work with Apps feature is now open to all users.

This feature allows users to read code and other data from text editing and IDE tools without repeatedly copying and pasting, making programming and other tasks more convenient.