Google is expected to announce a new Gemini model at its annual I/O conference this coming Tuesday, May 19, 2026. Current reports suggest the release may feature Gemini Omni, a video-generation model in testing, with performance capabilities positioned alongside OpenAI’s GPT-5.5.
Gemini Omni and the Shift Toward Video Integration
The upcoming Google I/O conference is expected to center on the expansion of multimodal capabilities, specifically through a model reportedly named Gemini Omni. According to reports from Gadgets360, this model is currently in testing and aims to allow users to generate and edit video content directly within the Gemini interface. This development appears to be an evolution of Google’s existing Veo technology.
The computational intensity of these new video features is already becoming apparent in preliminary tests. Reports indicate that just two test generations of video used 86 percent of a daily AI Pro quota
. This high resource consumption highlights the technical challenges of integrating high-fidelity video generation into consumer-facing AI agents, as the hardware requirements for video processing exceed those of text or image-based models.
For more on this story, see Google Cloud partners with CVC, GlobalLogic, and Tredence to overcome enterprise AI integration barriers.
Competitive Benchmarking and the I/O 2026 Reveal
While the industry anticipates a significant update, the new model may not represent a total leap in frontier capabilities. Reporting from Alex Heath suggests that the upcoming Gemini release is expected to land in the same class as OpenAI’s recent GPT-5.5. However, the model is not expected to set a new industry ceiling; Heath noted that the release will be well short of Anthropic’s Mythos
, which has reportedly reset the standard for what constitutes leading AI performance.
This positioning suggests Google is focused on parity and functional integration rather than attempting to leapfrog the current state-of-the-art in pure reasoning. By targeting the GPT-5.5 tier, Google appears to be prioritizing the deployment of reliable, multimodal tools that can be integrated into its existing ecosystem of apps and services.
The Rapid Iteration of the Gemini Ecosystem
The move toward Gemini Omni follows a period of rapid deployment for Google’s AI division. The company released Gemini 3 on November 18, 2025, which introduced Gemini 3 Pro and enhanced multimodal reasoning. That release also included the development of Gemini 3 Deep Think mode, aimed at complex problem-solving for Ultra subscribers.
The transition from the Gemini 3 architecture to the potential Gemini Omni model demonstrates a strategy of continuous updates. By building on the foundations of Gemini 3 and Veo, Google is attempting to move beyond text-based interaction toward a more cohesive experience where video and reasoning are handled within a single, unified model architecture. The industry will watch Tuesday’s announcements to see if Gemini Omni can bridge the gap between current capabilities and the high-performance standards set by Anthropic.
