ChatGPT and Gemini: what’s new for the two rivals?

This start of the week promises to be intense in the world of tech! If Google had planned its I/O conference well in advance on Tuesday, May 14, OpenAI came to steal the show by organizing a live event this Monday, May 13 at 7 p.m. (French time) to unveil its latest updates for ChatGPT and GPT-4. Overview of the latest news from OpenAI and Google, as well as a small overview of the latest Microsoft announcements!

What’s new for OpenAI and ChatGPT

With its 1.77 billion visits per month, ChatGPT still dominates the conversational AI market, and easily outdistances its main competitors such as Gemini, Character AI, Perplexity and Claude, according to data from Similarweb (study). The latest new features to remember are as follows:

  1. OpenAI has just unveiled GPT-4o, a new freely available AI model that enhances ChatGPT’s text, voice and vision capabilities.
  2. GPT-4 Turbo is the best performing AI model according to Chatbot Arena.
  3. ChatGPT has been accessible without registration since the beginning of April.
  4. ChatGPT has a new editor to modify your AI-generated images.
  5. OpenAI launches tool to detect content generated by DALL-E 3. Its effectiveness still remains to be proven. As a reminder, OpenAI launched a tool to detect text generated by ChatGPT, then resigned itself to deleting it.
  6. OpenAI is working on its new GPT-5 AI model and plans a release for this summer. GPT-5 could integrate major advances: better performance, video format support, fewer hallucinations, more customization.

At the same time, OpenAI continues to work on Sora, its AI video generator. First sample videos were presented, and a music video was entirely generated by AI and uploaded to YouTube. Sora could be available before the end of the year according to statements by Mira Murati, the technical director of OpenAI, in an interview with Wall Street Journal.

What’s new for Google Gemini

Google continues to develop its star of the moment: Gemini, formerly called Bard, and main rival of ChatGPT. Gemini is accessible free of charge via web browser. A mobile application exists but it is only available in the United States at the moment. The latest new features to remember are as follows:

  1. Gemini is the 4th best performing language model according to Chatbot Arena, behind GPT-4 Turbo, GPT-4 and Claude Opus 3.
  2. Gemini exists in a free and paid version (Gemini Advanced). The paid version is only available in the United States for now.
  3. Google has just added extensions to Gemini to connect the main applications: Gmail, YouTube, Google Maps, etc. The objective is to facilitate access to information for users.
  4. Google offers a guide to prompts for Gemini “to write effective prompts to boost your productivity”we can read on the site.

What about Microsoft?

Microsoft is also continuing its AI race. At the beginning of May, the American firm notably unveiled Phi-3, small AI models. Another new thing to remember: at the end of April, Microsoft presented VASA-1, an AI model capable of generating “speaking virtual characters” from a simple photograph and an audio track. At the same time, the firm also continues to highlight Microsoft Copilot, its assistant powered by GPT-4, which is now available via a dedicated interface (outside of Bing) and via a mobile application (Android, iOS).