Over 700 million people globally now use generative AI every week on ChatGPT alone, and OpenAI plans to hit 1 billion ChatGPT users by the end of 2025.

Generative AI refers to systems powered by large language models that generate content, such as text, code, or images, based on user input.

ChatGPT, the original AI chatbot released in November 2022, captured roughly 80% of the AI market and quickly became the fastest-growing online service in history. But are people using ChatGPT because it's better, or because of its recognizable brand name and brilliant marketing?

Overchat AI, a service that provides easy access to all major AI models on a single platform, has analyzed usage data from 28 million AI interactions to answer this question. Some of the results are surprising.

What AI Models Are People Choosing?

LLM technology isn't unique to ChatGPT. Competitors including DeepSeek, Mistral, Anthropic, Alibaba, and others promise smarter responses for less.

So when marketing power is removed from the equation and users are given equal access to AI models, do they still prefer GPT models, or do they gravitate toward other developers?

When the models are given a level playing field, most people don't choose GPT-5, OpenAI's latest LLM. Instead, users prefer open-source models from Chinese developers.

Data from Overchat AI, shows that roughly 80% of users interact with Chinese models.

Notably, this is not because of the user-base demographic, or marketing:

"Over 60% of our users are English-speaking, mainly based in the US, but despite that they seem to just trust the Chinese models more," says Andrew Dyuzhov, CEO of Overchat AI. "This is based on millions of AI messages our users send every week."

According to Dyuzhov, Chinese models don't have any special marketing or increased visibility in the in-app store — people naturally pick them more often, suggesting that performance, not brand recognition, drives user choice when barriers are removed.

This analysis is based on 28 million AI messages processed through Overchat AI's platform over the last 3 months (July 2025 – September 2025).

The platform provides unified access to multiple AI models, allowing users to freely choose between options without price differences or promotional bias. The user base comprises over 60% English-speaking users, primarily from the United States, with the remainder distributed across Europe, Latin America, and Asia. Users can switch between models at any time, providing real-world data on preference patterns when given equal access to competing AI systems.

Note that AI usage trends may vary by platform.

Brands Vs. Tools: What Really Matters to Users?

When given equal choice, do people gravitate toward branded models (GPT-5, DeepSeek, or Flux.1 Kontext for images) that they recognize, or do they prefer specialized AI tools (AI Writer, Text Paraphraser, Image Generator) that promise specific functionality?

• People using text-based chatbots are just as happy to use productivity tools, even when they don't know which model powers them.

• However, with visual models, people tend to prefer recognizable brands like Veo3, Nano Banana, Seedream, and Qwen Image.

What are people using AI for?

Most consumers use AI to generate images and summarize long texts and documents, according to Dyuzhov. Only about 20% of people use chatbots to have multi-turn conversations.

What Tasks are Users Solving with AI Most Often?

Diving deeper into specific natural-language tasks, we looked at the recent NBER research based on OpenAI's data (Overchat AI doesn’t collect chat data).

• 77% of requests are for advice, information searches, and copywriting. Interestingly, only a third of the copywriting requests are for original content — most involve editing or reformatting existing text

• 10% of requests are learning-related

• 4% are coding-related

However, the type of questions users ask AI heavily depends on the platform and its user base. Claude, which has cultivated more enterprise and business clients, sees that 37% of requests are code-related and just 10% for general text.

How has AI Usage Change Over the Years?

The demographic of AI tools seems to be shifting away from power users and towards everyday people, who use AI AI to solve simple tasks:

• The volume of AI users has increased fivefold in just one year. The adoption is faster in underdeveloped economies.

• While 80% of ChatGPT users were male after its release, the gender split is now 50/50.

• About 46% of users are between 18 and 25 years old.

Future Outlook

When given equal access to various models, users seem to prioritize performance over brand recognition, with 80% of users on Overchat AI opting for open-source models such as DeepSeek and Qwen over GPT-5, for text-based tasks at least.

When the aim is generating a video, in contrast, users tend to gravitate towards recognizable brands.

On platforms that target individual customers as opposed to businesses, cCoding related requests comprise an overwhelming minority — under 4% of the total. As the number of everyday people using AI models is increasing, developers will likely pursue that demographic to increase their market share.

While this may sound like bad news to power users, developers will likely continue to fine-tune top models to provide better responses in exchange for simpler prompts.

###

Media Contact

Katerina

Overchat AI

hello@overchat.ai

https://overchat.ai/

newsroom: news.38digitalmarket.com

Source published by Submit Press Release >> How do People Use AI? New 2025 Usage Statistics