Last reviewed
May 9, 2026
Sources
No citations yet
Review status
Needs citations
Revision
v2 ยท 2,493 words
Add missing citations, update stale details, or suggest a clearer explanation.
Data analysis ChatGPT plugins were a category of third-party tools that extended ChatGPT with capabilities for analysts, data scientists, and finance professionals working with structured data in a chat interface. Active from late March 2023 until the plugin platform shut down on April 9, 2024, they let the chatbot ingest CSV files, query market and reference databases, generate charts, and run notebook-style code on a user's behalf. Together with OpenAI's own Code Interpreter feature, they formed the early production toolkit for natural-language data analysis before being supplanted by Custom GPTs in the GPT Store.
This article is a historical reference. The plugins listed here are no longer reachable through the original interface; many developers later released equivalent functionality as Custom GPTs, standalone web products, or Model Context Protocol servers. Code Interpreter, sometimes conflated with this category, was a first-party OpenAI beta tool rather than a third-party plugin, but is discussed below because the two surfaces overlapped heavily in practice.
OpenAI announced the ChatGPT plugin system on March 23, 2023, alongside twelve initial partners (Expedia, FiscalNote, Instacart, KAYAK, Klarna, Milo, OpenTable, Shopify, Slack, Speak, Wolfram, and Zapier) plus first-party browsing and code execution plugins.[1][2] Of those twelve, Wolfram was the only launch partner with a strong data-analysis posture, but the platform was open: any developer could publish a plugin by exposing an OpenAPI manifest. Within weeks the store had filled with tools targeting CSV uploads, charting, financial data, and Jupyter integration.
Key plugin platform dates relevant to the data analysis category:
| Date | Event |
|---|---|
| March 23, 2023 | OpenAI announces ChatGPT plugins; Wolfram among 12 launch partners[1][2] |
| Late April 2023 | Roughly 70 plugins available; alpha access expands beyond the waitlist[3] |
| May 11, 2023 | Noteable announces its plugin connecting ChatGPT to hosted Jupyter notebooks[4] |
| May 12, 2023 | Plugin store opens broadly to ChatGPT Plus users[5] |
| June 2023 | Polygon.io publishes the Polygon ChatGPT plugin for stocks, options, forex, and crypto data[6] |
| July 6 to 11, 2023 | OpenAI promotes Code Interpreter from alpha to beta for all Plus users, providing a sandboxed Python environment[7][8] |
| Late August 2023 | OpenAI renames Code Interpreter to Advanced Data Analysis with no functional change[9] |
| November 6, 2023 | OpenAI DevDay introduces GPTs as a successor framework[10] |
| December 2023 | Noteable ceases operations, removing the plugin from the store[11] |
| January 10, 2024 | GPT Store opens[12] |
| March 19, 2024 | Plugin store closes; new plugin conversations disabled[13] |
| April 9, 2024 | All remaining plugin conversations end; platform fully shuts down[13] |
The data analysis category was always informal. OpenAI surfaced plugins by tags and curated lists rather than a strict taxonomy, so the same tool sometimes appeared under "Data," "Productivity," or "Research" depending on the week. For an overview, see chatgpt plugin categories.
The large language model behind ChatGPT in 2023 (then GPT-3.5 and GPT-4) had two known weaknesses for analytical work. It could not reliably do multi-step arithmetic, and it could not actually inspect a dataset; it could only describe one from what the user pasted in. Data analysis plugins addressed both by routing real data and real computation through a tool call.
Typical use cases included:
A chat session could host up to three plugins at once, encouraging combinations such as Polygon plus Wolfram plus a charting plugin for a "fetch, compute, visualize" loop. ChatGPT decided which plugin to call, formatted a JSON request against the plugin's OpenAPI schema, received the response, and incorporated it into its reply.
The table below lists plugins that published verifiable launch information and were widely covered as data analysis tools during the plugin era. Plugins that could not be confirmed through at least two credible sources are omitted.
| Plugin | Function | Developer | Verified period |
|---|---|---|---|
| Wolfram | Symbolic math, numerical computation, curated scientific and reference data via Wolfram Language and Wolfram|Alpha | Wolfram Research | March 23, 2023 launch[1][14] |
| Noteable | Generate and run Jupyter-style notebooks (Python, SQL, Markdown) from prompts in a hosted environment | Noteable | Announced May 11, 2023; delisted December 2023[4][11][15] |
| Polygon (later Massive) | Market data, news, fundamentals for stocks, options, forex, and crypto | Polygon.io | Plugin published June 2023[6][16] |
| Show Me Diagrams | Generate Mermaid, GraphViz, PlantUML, and similar diagrams inline; popular for flow and process charts | bra1nDump (open source) | Active in plugin store during 2023[17][18] |
| daigr.am | Generate bar charts, pie charts, line graphs, donut charts, and flowcharts from chat prompts | daigr.am team | Active in plugin store during 2023[19] |
| Mermaid Chart | Generate Mermaid diagrams inside ChatGPT and open them in the Mermaid Chart editor for further editing and team sharing | Mermaid Chart | Active in plugin store from 2023[20] |
The Wolfram plugin was the headline data-analysis-leaning launch partner. Stephen Wolfram framed it as giving ChatGPT "computational superpowers," letting the model formulate a query, send it to Wolfram|Alpha or the Wolfram Language kernel, and describe the result in natural language.[14] For users working on statistics, finance, physics, chemistry, or pure math, it corrected the weakness of GPT-4 on multi-step arithmetic and provided curated datasets covering chemistry, geography, astronomy, and finance. The plugin also exposed plotting primitives, so a user could request a function plot or histogram and receive an inline image generated by the Wolfram kernel.
Noteable was widely considered the strongest data analysis plugin of the era. Announced on May 11, 2023, it distributed a hosted notebook environment that the plugin could write to.[4] A user could ask ChatGPT to load a CSV, build pandas dataframes, train a small scikit-learn model, or plot results, and ChatGPT would author the cells and execute them in the user's Noteable workspace.[15] The plugin supported Python, SQL, and Markdown cells and produced a sharable notebook URL after each session. It was often preferred over OpenAI's own Code Interpreter for persistence, sharable URLs, and SQL connectors against external warehouses.[7] Noteable as a company ceased operations in December 2023, so the plugin disappeared from the store months before the April 2024 final close.[11]
The Polygon plugin connected ChatGPT to Polygon.io, a financial data provider covering U.S. and international equities, options, foreign exchange, and crypto.[6] A user could ask for the latest price action on a ticker, request aggregate bars for a date range, pull fundamentals from income statements and balance sheets, query historical dividends, or list options contracts for an underlying.[16] Because it returned real numerical data rather than model estimates, the plugin was useful for research workflows where accuracy mattered. Polygon.io later rebranded as Massive.com, and equivalent functionality moved to a Custom GPT after the plugin platform closed.
Show Me Diagrams, an open-source plugin by GitHub user bra1nDump, let ChatGPT emit and render diagrams in formats including Mermaid, GraphViz, PlantUML, Excalidraw, and Vega-Lite, using the Kroki rendering service.[17][18] Common uses included sequence and state diagrams, Gantt charts, pie charts, mind maps, and timelines.
daigr.am focused exclusively on chart generation, accepting a prompt describing the chart and returning a rendered image, with support for bar charts, pie charts, line charts, donut charts, and flowcharts.[19] It was a smaller, more focused alternative to Show Me Diagrams.
The Mermaid Chart plugin connected ChatGPT to the commercial Mermaid Chart editor.[20] A user could request a flowchart, sequence diagram, mind map, state diagram, Gantt chart, Sankey diagram, entity-relationship diagram, quadrant chart, or XY chart, then open the result in the Mermaid Chart workspace for editing and team collaboration.
Code Interpreter was an OpenAI-built feature, not a third-party plugin, but it shipped through the same plugin selector during the alpha phase. It was announced as part of the March 23, 2023 plugin launch and made generally available to ChatGPT Plus users in beta on July 6, 2023.[1][7] In late August 2023, OpenAI renamed the feature to Advanced Data Analysis with no functional change, signalling a focus on mainstream data work over coding.[9]
It offered a sandboxed Python interpreter with disk space, file uploads and downloads, and common scientific Python libraries.[8] A user could upload a CSV, Excel file, image, or zipped folder, and ChatGPT would author and run Python to clean, summarize, plot, or transform it. Third-party plugins in this category overlapped with Code Interpreter on at least one axis:
| Plugin | Overlap with Code Interpreter | Distinct value |
|---|---|---|
| Wolfram | Both can do math and statistics | Curated knowledge graph and symbolic math beyond Python's reach |
| Noteable | Both run Python | Persistent shareable notebooks, SQL connectors, longer-running kernels |
| Polygon | Both can ingest data files | Live market data without leaving the chat |
| Show Me Diagrams / daigr.am / Mermaid Chart | Both can produce charts | Diagram-specific syntax (Mermaid, GraphViz) for flows and architectures |
In practice many users mixed the two surfaces, opening Code Interpreter for CSV cleaning, then enabling Wolfram for a precise computation and Show Me Diagrams for a flow chart of the result.
Common plugin combinations followed a fetch, compute, visualize pattern. An analyst asked Polygon for the last twelve months of daily closing prices for a ticker, asked Wolfram to compute a rolling thirty-day standard deviation and an annualized volatility figure, asked daigr.am or Show Me Diagrams to draw a line chart of the price series and a bar chart of monthly returns, then asked Noteable to write a Python notebook reproducing the calculation for later reuse, sharing the resulting notebook URL with colleagues.
This pattern foreshadowed the agentic data tools that became common after 2024. It also pushed users into deliberate prompt engineering habits: explicit instructions to cite data sources, to refuse if no live data was returned, to surface tool errors rather than guess, and to break complex requests into one tool call per step.
OpenAI announced the wind-down of plugins on March 7, 2024, citing better developer ergonomics and a clearer commercial path under GPTs.[13] Factors that affected the data analysis category included poor plugin discovery (a flat store list with no ratings), the binding three-plugins-per-chat limit for analytical work that needed fetch plus compute plus visualization plus notebook, native coverage of file-upload-and-analyze use cases by Code Interpreter / Advanced Data Analysis, the operational burden of hosting an external server and OpenAPI manifest, and OpenAI's strategic preference for consolidating around the GPT Store as a future revenue-sharing marketplace.[12] Noteable's company-level shutdown in December 2023 also signalled that plugin-only distribution was hard to monetize.[11] Users had roughly three weeks of notice between the March 19, 2024 freeze and the April 9, 2024 final shutdown.[13]
Data analysis workflows that once ran on plugins migrated to four kinds of tools:
The data analysis category overlapped with several others on the plugin platform. Academic research plugins (ScholarAI, AskYourPDF, Consensus) shared workflows with data analysis plugins because researchers needed to extract numerical results from papers and recompute them. Productivity plugins (Zapier, Slack) were used to push computed results into trackers, sheets, and chat channels. Browsing-style plugins (Link Reader, WebPilot) were used to fetch data from public web pages not exposed through a dedicated finance or science API. For the broader history of the platform, see chatgpt plugins; for the live-era category map, see chatgpt plugin categories.
See also: ChatGPT Plugins, ChatGPT Plugin Categories and Data Analysis