Technology giant Alphabet hosted the keynote for its 2026 I/O developer conference on Tuesday, introducing a series of significant updates centered on "agentic AI," including new AI models, expanded Gemini application features, and a new pricing structure.
The Gemini 3.5 Flash and Omni models were unveiled. Alphabet released the next-generation AI model Gemini 3.5 Flash, which is now the default model for the Gemini app and Search's AI mode. The company claims this model surpasses the previous 3.1 Pro in coding and agent task benchmarks, operates at four times the speed of other leading models, and costs less than half. A more powerful Gemini 3.5 Pro version is expected next month. Additionally, the new multimodal model Gemini Omni can accept text, image, audio, and video inputs to generate video content, available to AI subscribers and YouTube Shorts creators.
A highlight is Gemini Spark, a cloud-based, always-on personal AI agent capable of performing tasks on a user's behalf. Spark is deeply integrated with Workspace apps like Gmail and Docs, enabling it to automatically parse credit card statements, monitor school email deadlines, and generate daily summaries. Alphabet also introduced an agent payment protocol, with plans for Spark to securely complete payment operations for users in the future. Spark opens to trusted testers this week, with a beta launch for U.S. AI Ultra subscribers scheduled for next week.
The Gemini app received a major redesign with a "neural expression" design language, featuring a more dynamic and interactive interface. A concurrently launched Daily Brief feature can generate personalized daily summaries based on a user's calendar and emails. Alphabet also adjusted its subscription pricing: a new entry-level tier for AI Ultra was added at $100 per month, while the original $250 tier was reduced to $200 per month. The billing model shifted from daily prompt limits to a "compute-based" model. The company revealed that Gemini has surpassed 900 million monthly active users.
Comments