40+ Surprising Google Gemini Statistics For 2025

Google Search has been getting worse, a trend noticed by many in 2024.

Will users turn to LLMs instead for more accurate information?

Launched in December of last year, Gemini is Google’s candidate in the arena of chatbots to contest those of the likes of Claude and ChatGPT.

Let’s have a look at the latest Google Gemini statistics.

Gemini User Adoption Trends

Generative AI is currently being adopted at a large scale all over the world.

These AI models are accessible online or even as applications on your phone such as the Gemini App.

Let’s look at how people are currently using Gemini right now, as well as the site’s visitor metrics.

Google Replaces Bard with Gemini

Google has replaced its old chatbot known as Bard with it’s new Gemini model, signaling a new era for Google. (Economic Times, 2024)

Google’s Gemini AI Assistant Mobile App Registered Around 783,632 Downloads

This was during the month of February 2024. (Statista, 2024)

62% of People Now Use ChatGPT or Google Gemini To Find A Product or Service

In an online poll conducted by Ignite Visibility, 62% of respondents said that they use either ChatGPT or Gemini to research products and services on the market. (Ignite Visibility, 2024)

Hence, Chatbots like these influence convincing a customer to buy a certain product.

Gemini gets more laptops/desktops users than on their phones, for more stats like this let’s look at the Similar Web:

Google Gemini Device Distribution: 60.09% Desktop Usage

39.91% use Gemini on mobile devices.

Google Gemini Visitor Metrics: 4:59 Average Visit Duration

The average user goes through 3.47 pages per visit.

Google Gemini Performance Metrics: 38.33% Bounce Rate

This is around the same value as ChatGPT, which has a bounce rate of 38.02%.

Stats on Gemini’s Global Language and Accessibility Reach

Gemini is widely accessible to anyone on the internet with a Google account, accounting for 40 languages.

It also has a significantly large amount of requests per minute compared to other large language models, as shown below.

Google Recently Increased The Free Version Of Gemini’s Rate Limit To 1,000 RPM

RPM stands for requests per minute. (Developers’ Google Blog, 2024)

Google Launches Gemini App on Play Store for Android Users

This app has a rating of 4.7 stars, along with more than 10 million downloads. (Google Play, 2024)

Gemini is Available in 40 Languages

For the mobile app though, Gemini is restricted by the supported languages of the mobile app. (Support Google, 2024)

It is Accessible in Over 230 Countries and Territories

You still require a Google and a supported browser, and be above the age of 13 to access Gemini. (Support Google, 2024)

Gemini Traffic and User Demographics

Gemini recently recorded more than 200 million visits from all over the world, yet its traffic has fluctuated.

Let’s look at its traffic stats along with its user demographics.

Google Gemini Records 274.7 Million Worldwide Visits in September 2024

This shows a marginal increase of 2.56% from the previous month. (Similar Web, 2024)

Google Gemini Experiences Steep Decline in Traffic in September

From 02 September to 23 September, traffic was steady at 60 million, yet after that, we saw a dip of over two-thirds with traffic dropping to 20 million. (Similar Web, 2024)

This trend can be seen with ChatGPT as well.

United States Leads With 14.06% Traffic Share

There are slightly more men than women visiting Gemini.

Gemini’s Visitors Are 58.52% Men

The other 41.48% comprise of women. (Similar Web, 2024)

Around 75% of those using Gemini are below 44 years of age.

31.10% of the Gemini Users Are Aged Between 25 to 34 Years

The second largest age group is the young adults (18-24) at 23.27% followed by those at 25-44 years old at 19.07. (Similar Web, 2024)

Facts Around Gemini’s Technical Capabilities

Let’s look at some features that Gemini has, that sets it apart from the other models.

For example, Gemini has the largest context window at 2 million tokens!

Let’s go into these, taking insights from Google Deepmind:

Google’s Gemini AI Arrives In Four Distinct Models

• Ultra: This is Gemini’s new model for ultra-complex tasks.

• Pro: When it comes to general performance, Pro is the one to go for.

• Flash: Flash is a lightweight model when speed is preferred.

• Nano: The model to use for on-device tasks.

Project Astra Explores The Future Of AI Assistants With Gemini-Powered Multimodal Agents

With this project, you can have conversations with your phone about multimodal information, such as the view from the camera, where the AI agent identifies objects.

Gemini Surpasses Other Models With Two Million Token Context Window

This content is equivalent to two-hour videos, 22 hours of audio, and around 1.4 million words.

Gemini Has Built-In Safety Classifiers To Filter Violent or Stereotypical Content

Yet Gemini Pro is still susceptible to jailbreak attempts.

According to research, the PAIR schematic for jailbreaks was an automated jailbreak that managed to circumvent Gemini Pro’s safety protocols. (Arxiv, 2024)

Gemini’s Architecture is Designed to Process Diverse Data Types

These data types include text, images, audio and video.

This is a key development in creating capable AI models that can do advanced data anlysis, as it changes how the machine interacts with the environment. (Arxiv, 2023)

Gemini’s Performance Benchmarks

Artificial Analysis conducted their evaluations on popular Large Language Models, and these are the results for Gemini.

Let’s get into these multimodal benchmarks.

Gemini Ranks Third in AI Quality Index

O1-preview and mini lead with a score of 85 and 82 respectively, followed by Gemini at 80.

Gemini Tops Speed Charts at 211 Tokens per Second

Llama 3.1 8B comes next at 168 tokens/ second.

Gemini Offers Competitive Pricing at $0.1 per 1M Tokens

It is the lowest price-to-token ratio of all the models.

Gemini’s is consistently strong across key competencies.

Gemini Ultra Scores 90% on the MMLU Benchmark

MMLU stands for Massive Multitask Language Understanding. (Arxiv, 2021)

Gemini Ultra Surpasses GPT-4 on Multilingual Math and Summarization

Gemini Ultra had a score of 79 on MGSM and 17.6 on XLsum, while GPT-4 scored 74.5 on MGSM. (Arxiv, 2021)

Gemini Pricing and Subscription Options

Gemini has quite a few different plans depending on use cases, here are their prices:

Gemini is Available for $19.99/Month Under the New AI Premium Google One Plan

This premium plan also has a 2-month free trial. (Google Blog, 2024)

Gemini Code Assist Standard is Priced at $22.80 per User/Month

Gemini Code Assist is available to try at no cost until November 8, 2024. (Google Cloud, 2024),

Gemini Code Assist Enterprise is Priced at $54 per User/Month

Gemini Code Assist Enterprise is available for $19 per month per user on a 12-month commitment until March 31, 2025. (Google Cloud, 2024),

Final Thoughts: What’s New for Gemini?

There are new features that Gemini users can look forward to in the future, according to a blog from Google:

• Recently NotebookLM- a PDF summarizer with an audio overview went viral as it generates podcast-like discussions from source material, this could be very useful for future research.
• The Gemini App Team will join Google DeepMind, with both teams focusing on improving feedback loops and deploying models quickly.
• User Experience will also be given priority for AI assistants.