AI Showdown: ChatGPT vs. Grok-3 vs. DeepSeek vs. Claude – Who Makes the Best Charts?
Hassan Syed
Architect | Cloud Advisor | Azure Certified Solution Expert | Generative AI | Enterprise Systems Expert | IoT Solutions | Big Data | Digital Transformation Leader | Integration Architect | Hands-on| Mentor
I recently tested ChatGPT, Grok-3, DeepSeek, and Claude on data visualisation by asking them to generate tables and graphs for a same prompt.
The prompt for the data table was to list the land area and population (in billions) for each continent and draw a chart.
The results were fascinating and I made different LLMs work together, see the details below. That showed the value of using multiple AI models with different strengths.
It shows that in the future, AI agents powered by different LLMs/SLMs will likely be working together like a team in a workplace, just as people do, each contributing their skills to achieve amazing results.
So here we go with the competition.
PS: Also check with my post on the AI Code Challenge https://www.dhirubhai.net/pulse/comparing-ai-generated-spaceflight-animations-grok-3-vs-hassan-syed-6z4kc/
?? Round 1: Table & Graph Generation
? ChatGPT: Created both a table and a visually smart graph, combining bars + a line graph in a single chart to compare multiple values effectively. Note: I used a gpt created by chatgpt here: https://chatgpt.com/g/g-HMNcP6w7d-data-analyst
The data source table from ChatGPT (the easy part, every llm did it right, so we will focus on the graphs only) :
?? ChatGPT’s Dual-Axis Graph:
? Claude: It successfully created the graph but split the data into two charts instead of one combined graph.
? Grok-3: Only generated a table, but provided a detailed textual description of the chart.
?? DeepSeek: Generated Python code but couldn’t render the graph.
?? Round 2: Using AI-Generated Descriptions to Render Graphs
Since Grok-3 and DeepSeek didn’t generate graphs, I fed their descriptions into ChatGPT to see if it could bring their ideas to life.
? Grok-3’s description rendered by ChatGPT: The result was well-structured and visually strong.
?? DeepSeek’s first attempt (Python code): The two graphs were basic and not very engaging.
?? Graph, Generated by ChatGPT, from DeepSeek’s Second Attempt (Based on Human-Style Description):
Graph, Generated by ChatGPT, from Claude AI’s Second Attempt (Based on Human-Style Description): It was identical to what it generated by itself in the first round.
?? Final Verdict: Which AI Did Best?
?#1. ChatGPT is best for both creating and rendering graphs.
? #2 Grok-3 & DeepSeek understand visualisation but need help to render. Upon Rendering both graphs came close to each other, so a tie.
? #3 Claude also rendered all in one go, however two separate graphs generated demonstrated lack of efficiency. When asked to write the description for a good looking graph the results were not much different.
Which AI do you use for data visualisation? Let’s discuss! ??
#AI #DataVisualisation #ChatGPT #Claude #Grok3 #DeepSeek #MachineLearning #ArtificialIntelligence
PS: Also check with my post on the AI Code Challenge https://www.dhirubhai.net/pulse/comparing-ai-generated-spaceflight-animations-grok-3-vs-hassan-syed-6z4kc/