Google Launches Gemini 3.5 Flash Model and Spark AI Agent Google used its I/O 2026 conference to argue that the future of AI is about autonomous agents working in the background, not just chatbots on a screen. But the launch of its new Gemini 3.5 Flash model and Spark AI agent landed alongside delays, access limits, and questions about who really benefits.

Timeline: From model delay to agentic pivot

On May 19, attendees at the packed I/O keynote learned that Google’s most powerful upcoming model, Gemini 3.5 Pro, wouldn’t ship that day after all, prompting audible disappointment. CEO Sundar Pichai told the crowd it would arrive “next month,” while stressing that it was already showing “great improvements” over earlier versions.

Instead, Google made Gemini 3.5 Flash the star. The company pitched Flash as its “strongest” model yet for coding and agentic tasks, with a balance of “speed and performance” suited to long-running workflows. TechCrunch reported that Flash can independently execute coding pipelines and even “build an operating system entirely from scratch,” marking a shift “from pitching AI as a conversational tool to AI as an agentic tool.”

Engineers amplified that message on social media. Chrome engineering lead Addy Osmani called Flash “fast, great for building rich UIs + agents and… stronger from coding to long-horizon tasks and multi-step workflows.” DeepMind CEO Demis Hassabis claimed it “performs better than 3.1 Pro on coding & agentic tasks,” while being up to “4x faster than other frontier models” and far cheaper. Pichai highlighted that Flash is “available today… across our products and APIs” and beats 3.1 Pro “across almost all benchmarks with huge progress in coding.”

Spark: An always‑on personal agent

Running on top of Flash, Google introduced Gemini Spark, its first “always-on” personal AI agent. Business Insider described Spark as a 24/7 assistant that can “plan parties, collate notes, and run a variety of other tasks in the background,” turning Gemini into “a more proactive helper.” Spark lives inside the Gemini app, runs in the cloud so “you don’t need to keep your laptop open,” and can pull from Gmail, Docs, Sheets and Slides to draft messages or organize information.

Google’s broader Gemini app is being rebuilt as an “all-purpose AI hub” with a new Neural Expressive design, a “Daily Brief” that compiles inboxes, calendars and tasks, and integrated voice via Gemini Live. The Verge noted that Gemini 3.5 Flash becomes the default model in Gemini and AI Mode, and is also “the model behind the new Gemini Spark AI agent for consumers.”

Enthusiasm, skepticism, and access limits

Coverage from Ars Technica emphasized Flash’s efficiency pitch: Google says it delivers “frontier-level intelligence” at nearly 300 tokens per second, potentially saving heavy enterprise users “a billion dollars per year” in compute if they switch from slower models.

But other observers questioned whether Google’s agentic push meets consumer needs. TechCrunch argued that, while Spark and related “information agents” could help track trends, inventories or travel plans, the demos—like using an agent to organize a neighborhood block party—felt needlessly complex compared to “a group chat or some emails.” Many of these features are also gated behind expensive Gemini Pro and Ultra subscriptions, including a $100‑per‑month Ultra tier, which could alienate average users.

Even on benchmarks, Gemini 3.5 Flash is facing external scrutiny. In response to new CursorBench coding results where Flash placed 10th, Elon Musk pointed users instead to a rival model, urging followers simply to “Try Composer 2.5.”

Amid applause from Google’s own leaders and mixed reviews from the broader ecosystem, the company’s bet is clear: make agents like Flash and Spark indispensable workhorses embedded across its products, while the more powerful—and still awaited—Gemini 3.5 Pro looms just over the horizon.

Continue reading https://foxvector.com