Google Launches Gemini 3.5 Portfolio, Introduces Agentic Framework and New AI Pricing

Google unveiled its newest AI model family during its annual developer conference, beginning with the Gemini 3.5 Flash model and with a Pro version scheduled for next month. [4] The Gemini 3.5 Flash model achieved 76.2% on the Terminal-Bench 2.1 coding test, surpassing the previous 3.1 Pro flagship model which scored 70.3%. [5] Google's public documentation notes that the Gemini 3.5 Flash model features a one-million-token context window alongside a maximum output limit of 65,000 tokens. [3] In addition to the Flash model, Google introduced Gemini Omni, which is capable of generating high-quality videos from text, images, audio, and other video inputs. [4] The Omni model includes multi-turn conversational editing capabilities and a built-in world model to simulate physical rules like gravity and fluid dynamics for applications such as physical mechanics demonstrations. [5]

To support developer operations, Google introduced Gemini 3.5 Flash (Low), a variant designed to reduce token consumption on simple tasks within the Google Antigravity platform. [3] This Low variant generates approximately 45% fewer tokens than the standard Flash release, which has been renamed to Flash (Medium). [3] The company also launched Antigravity 2.0, which supports multi-agent parallel execution, and Managed Agents, which allow developers to launch isolated Linux environments via API calls. [5] Alongside these infrastructure updates, Google introduced a $100 per month AI Ultra plan aimed at developers and advanced creators. [4] The tech giant is also lowering the price of its top-tier AI Ultra plan from $250 to $200, while offering an entry-level AI Plus plan starting at $7.99. [4]

These underlying AI capabilities are now impacting Google's consumer products, including a May 2026 Core Update that integrates Gemini 3.5 Flash directly into Google Search. [1] The algorithm update incorporates AI-generated search answers and multimodal improvements, resulting in reported traffic and ranking fluctuations across various industries over the rollout weekend. [1][2] Additionally, Google released Gemini Spark, an AI agent that operates continuously on Google Cloud virtual machines without relying on a local desktop or mobile device. [5] The Gemini Spark agent is designed to automatically track tasks, send follow-up emails, and pause at key milestones to await user confirmation across Gmail, Calendar, and Google Docs. [5]