OpenAI recognized three University of Pennsylvania student projects with its inaugural ChatGPT Futures award in May 2026, each receiving $10,000 and access to cutting-edge technologies. The awards highlight how students are leveraging ChatGPT to solve real-world problems, from accessibility for blind gamers to housing support for vulnerable families. Meanwhile, new market data reveals that ChatGPT's dominance in the AI chatbot space is gradually eroding, even as the platform maintains commanding reach with over 100 million users and 1.5 billion monthly website visitors.

What Are Students Building With ChatGPT?

The three Penn award recipients represent a diverse range of applications. Rising Wharton and Engineering sophomore Crystal Yang developed Audemy, a gaming platform designed specifically for blind students. Yang's platform has grown to include 50 games accessed by 200,000 individuals since she built her first game in high school. She explained that ChatGPT played a crucial role in her development process, from creating prototypes to refining outreach communications.

Rising Engineering senior William Sanz and rising Wharton seniors Praja Tickoo and Jack Patel created CaseLink, an artificial intelligence tool that helps case managers develop personalized action plans for families experiencing homelessness or housing instability. The platform allows case managers to work more efficiently and support more families with timely assistance.

"With the new API credits, we'll be able to test more and refine our model to make sure that the information is as relevant and accurate as it can be," said William Sanz, one of CaseLink's developers.

William Sanz, Rising Engineering Senior at University of Pennsylvania

Rising Engineering senior Parsa Idehpour received recognition for BioReason Pro, a tool that helps users understand protein function. Idehpour and his high school collaborator trained the model by inputting data into ChatGPT to create readable, sensible outputs. With OpenAI's API credits and financial backing, the team is now scaling their model training significantly.

"Our inspiration was in building AI scientists who can think better than humans and have superhuman capabilities. They can help us cure diseases and understand biology better," explained Parsa Idehpour.

Parsa Idehpour, Rising Engineering Senior at University of Pennsylvania

Why Did These Students Choose ChatGPT Over Competitors?

When asked why he selected ChatGPT instead of alternative AI models, Idehpour pointed to performance and reliability. He noted that ChatGPT had been "outperforming" other options, particularly in terms of "reliability" and "having the knowledge to do things accurately". This preference reflects a broader trend among developers who value consistency and accuracy in AI systems.

The ChatGPT Futures award program itself is noteworthy. OpenAI granted 26 prizes total to honor the Class of 2026, which the organization describes as the "first graduating class to have ChatGPT on hand for their entire university experience." Recipients gain not only financial support but also access to OpenAI's latest technologies and API credits, which are prepaid virtual currency used to access coding agents and other advanced features.

How to Leverage ChatGPT for Student Innovation Projects

• Prototype Development: Use ChatGPT to rapidly build initial versions of your idea, from game mechanics to application logic, without requiring extensive coding knowledge upfront.
• Content and Communication: Leverage ChatGPT to draft and refine outreach emails, documentation, and user-facing copy, saving time on writing and editing tasks.
• Data Processing and Model Training: Input raw data into ChatGPT to transform it into structured, readable formats suitable for training machine learning models at scale.
• API Integration: Access OpenAI's API credits to build production-grade applications that can handle real-world usage patterns and user loads.

Is ChatGPT Losing Ground in the AI Market?

Despite its continued dominance, ChatGPT's market share has declined modestly over the past eight months. In May 2025, ChatGPT held 74.9% of the AI chatbot market. By April 2026, that figure had dropped to 73.1%, representing a 1.8 percentage point decline. The decline was not linear; ChatGPT peaked at 75.1% in October 2025 before gradually sliding through early 2026, hitting its lowest point of 73.0% in March 2026.

Competitors are gaining traction. Google Gemini has steadily increased its market share from 13.4% in May 2025 to 15.1% by April 2026. Perplexity and ClaudeAI have also captured growing portions of the market, with ClaudeAI reaching 5.0% market share by April 2026, its highest recorded level. This fragmentation suggests that while ChatGPT remains the market leader by a significant margin, users are increasingly exploring alternative AI platforms.

OpenAI's financial performance, however, remains robust. The company is currently earning $2 billion in monthly revenue, with 2025 annual revenue reaching $13.1 billion. At the end of February 2026, OpenAI's annualized run rate exceeded $25 billion. The company has set an ambitious goal to reach $125 billion in revenue by 2029.

What Do ChatGPT Users Actually Do With the Platform?

Usage patterns reveal that general research dominates ChatGPT activity. In May 2025, general research accounted for 36.4% of all ChatGPT usage, followed by academic research at 18.6%, coding assistance at 14.1%, and email composition at 14.0%. These proportions remained relatively stable through April 2026, with general research consistently hovering around 36.5% to 36.8% of total usage.

The demographic breakdown shows that 55.99% of ChatGPT users are men, while 44.01% are women. When handling personal tasks, 83.27% of individuals opt for ChatGPT, indicating strong user preference for the platform in everyday applications.

ChatGPT's pricing structure has evolved since its launch. The platform was free until February 2023, when OpenAI introduced ChatGPT Plus at $20 per month. Plus subscribers gain access to advanced features including DALL-E image generation, web browsing, advanced data analysis, and GPT-4 access. The free tier uses GPT-3.5, while Plus users can access the more capable GPT-4 model, which can generate up to 25,000 words and understand more than 26 languages.

The investment required to build these models is substantial. GPT-3 cost OpenAI approximately $12 million for a single training session, while GPT-4 training incurred expenses of roughly $100 million. Despite these costs, the platform has achieved rapid adoption; ChatGPT attracted 1 million users within five days of launch, and its iOS application was downloaded 5 million times within three weeks of release.