Google unveils Gemini 3.5 Flash, Omni, and Spark AI
What's the story
At the Google I/O 2026 event, the tech giant has unveiled a slew of updates to its Gemini model family. The most notable addition is Gemini 3.5 Flash, a new model that promises to be faster and more capable than its predecessors. Along with this, Google also introduced Gemini Spark and Omni, two innovative AI tools designed to enhance user experience across various platforms.
Model evolution
Gemini 3.5 Flash
Gemini 3.5 Flash is the latest addition to Google's model family, and it promises a significant leap in building more capable and intelligent agents. The new model will be the default across several Google services, offering faster response times without compromising on intelligence. It also outperforms its predecessor, Gemini 3.1 Pro, on coding and agentic benchmarks while leading in multimodal understanding capabilities.
Enhanced features
New model can deploy teams of subagents
Gemini 3.5 Flash is designed to handle "long-horizon agentic tasks" and deploy teams of subagents. It can also create "richer, more interactive web UIs and graphics." The model is already available for users through the Gemini app and AI Mode in Google Search. A Pro version of this model is expected to be launched next month, further enhancing its capabilities.
Video innovation
Gemini Omni generates videos from any input
Along with Gemini 3.5 Flash, Google also launched Gemini Omni, a multimodal model for generating polished videos from any input. Unlike its predecessor Veo 3, which focused on converting text into video, Omni can take any combination of images, audio, video and text to create videos "grounded in Gemini's real-world knowledge." The resulting videos can be easily edited through conversation.
Editing features
Potential misuse in video editing
Gemini Omni can be used to create visual explainers for complex topics, with the ability to edit using natural language. However, this could also lead to potential misuse such as surreptitious editing of videos. Google is taking a cautious approach in deploying Omni, testing the ability to edit videos responsibly while ensuring each generated video carries a SynthID digital watermark for authenticity verification.
AI assistant
Meet Gemini Spark, your personal AI agent
With the launch of Gemini 3.5 Flash, Google is also introducing Gemini Spark, a personal AI agent. Unlike its predecessor that just answered questions, Spark acts as an "active partner that does real work on your behalf and under your direction." It can automate tasks like parsing monthly credit card statements for hidden fees or new subscriptions and monitoring inboxes for updates from schools.
Expansion strategy
Gemini Spark to integrate with Google Workspace apps
Gemini Spark will be integrated with Google Workspace apps like Gmail, Docs, and Slides. It will also work with Gemini-connected apps such as Canva, OpenTable, and Instacart. Google has big plans for Spark's future capabilities including texting/emailing with it, creating custom sub-agents, and operating a local browser. The AI agent is rolling out to trusted testers this week before its beta launch next week for US-based Google AI Ultra subscribers.