ChatGPT versus Gemini: the battle for users in the AI space

Artificial Intelligence has moved into a new phase after the public launch of Generative AI capabilities. ChatGPT was the app that started the torrent of releases, with increasingly specialised apps being designed for specific tasks. OpenAI, the company that launched ChatGPT, found its competition in Google, which initially launched a software called Bard, later renamed to Gemini. We’re a long way from seeing AI at full capacity, but we’re clearly witnessing an evolution—one that remains hotly debated regarding its potential benefits and drawbacks.

By Aurel Dragan

Many voices in the public space, some with a science background, are pointing to the unacknowledged risks of large-scale AI use. Unfortunately, no matter how right they may be, the use of Artificial Intelligence started a long time ago, specialising in Machine Learning, automating increasingly complex repetitive processes, and taking over jobs that could be done digitally, such as chatbots. As data processing capacity has grown, so has the analytical capacity of AI engines.

ChatGPT was launched in November 2022 by OpenAI and currently has 100 million active users, which can interact with it online on the OpenAI website. The free version allows text input, while the paid versions also support images and voice. The standard version of ChatGPT today is a well-crafted OpenAI GPT 3.5 LLM (Large Language Model) edition, which uses various technologies such as machine learning, NLP (Neuro-linguistic programming), and NLG (Natural language generation) to process user requests. ChatGPT offers those willing to pay two more powerful versions, GPT-4 and GPT-4o, where “o” stands for “omni”. An important difference between the free and the paid version is the limitation on the number of queries that can be submitted.

It is also important to note that the free version of ChatGPT is limited to the level of data existing in April 2023 and has no knowledge of events occurring after this date. It will probably get an update when a new version is released. ChatGPT-4o can also be used for free under certain conditions, with the data used for its training stopping in October 2023. ChatGPT Plus doesn’t limit the number of queries, but users need to pay at least USD 20 per month for this convenience.

ChatGPT 4 Turbo is capable of processing up to 300 pages of text for a single query and easily creates advertising messages or literary texts, poems, articles, and programming code. It can describe images and translate text from 95 languages. Its main competitor, Gemini, also creates, analyses, and translates various text and code formats in over 40 languages, and the multimodal neural network is capable of generating images based on text queries or other images.

Google’s AI chatbot

Gemini, formerly Bard, was launched in December 2023 by Google DeepMind, Google’s AI development division. The late launch compared to competitors has allowed it to develop a powerful multi-purpose AI service, a universal intelligent bot that works with text, code, images, video, and audio. Its abilities allow it to create, analyse, edit, summarise, and combine content. It can be used on almost any device, from smartphones to data centres, which is a convenient feature.

Gemini 1.0 is available in three versions: Nano, Pro, and Advanced, with paid versions starting at USD 7 per month. The first is the smallest in size and requires few resources, allowing it to be used on mobile devices. The Pro variant comes with an optimal balance of power and resource intensity, making it suitable for a wide range of tasks, while the full-featured Gemini Advanced version offers maximum capabilities and is used for the most complex and resource-intensive processes.

In February 2024, DeepMind released a new version of its AI bot, Gemini 1.5 Pro. Its main enhancement focused on context understanding. The updated model processes up to 1 million data inputs (tokens), allowing it to analyse large texts, videos, and code snippets. It’s currently available to authorised developers and the company’s corporate customers through the Google AI Studio and Vertex AI platforms. In the future, Google plans to open Gemini 1.5 Pro to regular users, but right now they can only use the Gemini 1.0 model line, including Gemini 1.0 Ultra with a paid Gemini Advanced subscription.

On May 14, 2024, another new product was unveiled at the Google I/O conference: Gemini 1.5 Flash. It’s a compact, multi-mode model designed to scale efficiently and handle specialised, high-frequency applications. It comes with a library of 2 million tokens and is already offered in public preview via the Gemini API on the Google AI Studio platform. In addition, Gemini 1.5 Pro, which debuted in February, has already received an update, and the database has also expanded to 2 million tokens. Gemini 1.5 Flash is performance-focused, making it ideal for fast, real-time client responses or on-the-go image generation. Gemini 1.5 Pro, with more power, is similar in functionality to Gemini 1.0 Ultra and is better suited for scientific text analysis and summarisation. Both models are capable of processing text, images, and videos. An advantage of the Gemini is also the online search enabled by access to Google services, a feature that is only available from ChatGPT on the paid version, while the Gemini offers the feature in all versions.

Self-learning capacity

One feature that both apps have is the extension of standard capabilities using add-ons. In January 2024, OpenAI introduced the GPT Store, a store of customised versions of ChatGPT available to those with subscriptions. Currently, there are over 3 million AI apps developed on ChatGPT by third parties. For convenience, these are divided into categories (text writing, programming, education, science, and others). Gemini also has its own app marketplace for integration with Google services and apps. Its add-ons resemble Google Chrome browser extensions, and with them, users can automate a range of tasks and processes. For example, they can create email newsletters and presentations, search for information in their inbox, receive AI recommendations for places or activities based on Google Maps data, and so on. Gemini add-ons are currently only available for Google Apps.

But the biggest difference between the two models currently is the self-learning capability. OpenAI learns independently as it communicates with users, understands context, and takes it into account in subsequent requests, whereas Gemini’s capabilities in this direction are still very limited. In the future, Google will likely solve this problem.

The two AI applications are the most comprehensive on the market. Gemini excels at user-experience, being easy to use, while ChatGPT focuses on companies and professionals who need a universal tool for content generation, data analysis, and customer interaction. Moreover, it can be integrated into many third-party systems.

But ChatGPT and Gemini are not alone on the market. Today, there are many AI software packages specialising in specific tasks. Among them, the best free ones include Jasper, which offers the best mix of image processing and copywriting; Scalenut, for very good SEO copywriting; Rytr, which has the most generous free plan; Writesonic, very good for writing articles; Simplified, good for workflow automation; Paragraf AI, with the best AI mobile app functionality; and CopyAI, the best choice for sales and marketing.