Screenshots published in early May on X reveal Gemini’s new interface design, already rolled out to select users. Behind this visual overhaul, Google is preparing its response to an increasingly competitive landscape against Claude and ChatGPT, less than two weeks before its I/O conference.
Key Takeaways
- Gemini’s new design, already visible to some users, features color waves and a fully redesigned interface
- Google I/O is scheduled for May 19, 2026, the likely moment for the official launch
- An omnimodal model codenamed Omni is reportedly in testing, potentially Gemini 3.2 or 3.5
A redesign that arrives before the conference
Before the conference even opens, Gemini’s new face has already started appearing for some users. Screenshots circulating since early May show a clear shift in aesthetic direction. The app now features color waves and a lighter interface, addressing the perception that the current version felt less modern than its direct competitors.
The redesign goes beyond a palette swap. The home screen has been reworked, along with the conversation list, the attachment menu, and the chat interface itself. It is a comprehensive overhaul of the user experience, designed to bring Gemini up to the level of ChatGPT and Claude, whose interfaces are consistently cited as more intuitive.
A partial rollout before the official announcement has become standard practice at Google. Testing in production with a limited group of users allows real feedback to be collected without waiting for the conference. The fact that screenshots leaked this early signals that the tester pool is already sizeable.
Google I/O on May 19, 2026 will likely be the moment of the official launch. Google has made a habit of using this event for its major consumer product announcements. A well-received redesign could restore visibility to a service that has been struggling to hold its position against the competition.
The timing is deliberate. In early May, market attention was largely focused on the releases of Claude Opus 4.7 and GPT-5.5. Arriving at Google I/O with a visible redesign and a potentially new model means reclaiming the initiative before summer.
Google under pressure since late 2025
In late 2025, Google made a strong move with Gemini 3 Pro, widely regarded at the time as the best model on the market. A few months later, the picture has shifted. Claude Opus 4.7 is dominating conversations among advanced users, and GPT-5.5 has put OpenAI back in a position of strength for the first time in several months.
This shift in dynamics is significant. Gemini 3 Pro had allowed Google to briefly take the lead in a benchmark race that had long played out between OpenAI and Anthropic. Losing that position is not permanent, but it illustrates just how fast the pace of progress is in this space.
Google has continued to push on other fronts. The launch of a Mac application for Gemini is a signal of expansion onto desktop platforms. Staying relevant is no longer just about model performance: the quality of native applications and the smoothness of the user experience are becoming real differentiating factors.
Google I/O arrives in a context where Google must respond on multiple fronts at once. More powerful models, better-designed apps, and a clearer product identity for Gemini against competitors who have built strong brand recognition. The stakes go well beyond the redesign alone.
Also on Horizon:
- iOS 27: Apple Plans to Let Users Pick Their Own AI Model
- Claude: how to stop wasting your tokens
- Apple Pays $250M to Settle Apple Intelligence Lawsuit
Gemini Omni: an omnimodal model in the works
Beyond the redesign, information from specialized testers suggests that at least one new model is currently being tested internally. This model, referred to as Omni, is said to be the successor to the Veo video generation model, with capabilities extended across all modalities.
An omnimodal model refers to a system capable of processing and generating text, images, audio, and video jointly, without relying on separate modules. If Omni matches this description, Google is firmly placing itself within a major industry trend: every leading player is pushing toward an all-in-one model that adapts to varied contexts without friction.
The name Omni also implies a clear lineage with the omnimodal approaches that gained traction in 2024. By adopting this positioning, Google signals that it is competing directly on this terrain against its main rivals.
The open question remains the scope of this model at launch. If it is a Gemini 3.5, the positioning relative to Gemini 3 Pro will need to be clearly defined. If it turns out to be a Gemini 3.2, it signals an interim update designed to maintain competitive pressure without waiting for a full overhaul.
Either way, Google I/O on May 19 will be decisive. Google has the resources and infrastructure to carry real weight in this competition. The question is not whether Gemini can return to the front of the race, but with what and at what pace.
Follow the story on Horizon.
Pingback: Anthropic x SpaceX: Claude Code doubles its limits - Horizon
Pingback: Google Is Building Remy, a Personal AI Agent 24/7 - Horizon