Vibe check: Comparing ChatGPT, Gemini, Claude & Plotly Studio
Meet the Speaker
Austin Kiesewetter
Austin Kiesewetter is a Plotly community member and software engineer at Martin Engineering.
This video walks through a head-to-head comparison of four AI tools for building interactive dashboards. ChatGPT, Gemini, and Claude are evaluated against Plotly Studio, with each system given the same CSV dataset and prompt. The goal was to see how general-purpose large language models stack up against a tool designed specifically for data science workflows.
The test highlighted several points.
- All three LLMs generated code, but none worked out of the box. They relied on outdated versions of Dash (2.18 instead of 3.0 or 4.0), required manual fixes for CSV encoding, and produced errors before a working app could be rendered.
- ChatGPT’s output used 172 lines of code and produced charts that were mislabeled and confusing.
- Gemini generated just two graphs, missing core elements like tables, and was fairly unsuccessful.
- Claude wrote over 550 lines of code, creating a more usable dashboard with filters and labeled states, but formatting issues and questionable labeling kept it from performing well.
In contrast, Plotly Studio accepted the same CSV through drag-and-drop, automatically structured the dashboard, and produced a polished, professional layout. It included collapsible filters, editable components, and accurate state-level views without manual bug fixing. While subscriptions to the other AIs cost between $20 and $200 a month, Plotly Studio offers its trial tier for free with optional paid upgrades.
Overall, the test shows the advantage of a purpose-built AI for data science tasks. Watch the video to follow along.
