At I/O, Google debuts Gemini 3.5 Flash for agentic and coding workflows

Google used its I/O 2026 keynote on May 19 to introduce Gemini 3.5 Flash, a new model it said is built for agentic workflows and coding-heavy use cases. The company described 3.5 Flash as the first release in a new Gemini 3.5 family and said the model combines “frontier intelligence with action.” Google tied the launch to a broader I/O push around agents, search and consumer products, including smart glasses and updates to the Gemini app. Google’s developer materials said the model is designed to handle real-world tasks that require speed as well as tool use. ### What exactly did Google announce at I/O? Google said at I/O 2026 that Gemini 3.5 Flash is the first model in its latest generation of systems for “more capable, intelligent agents.” In its main Gemini 3.5 post, the company said it was “kicking off the series” with 3.5 Flash rather than a Pro release. Google’s I/O roundup separately listed Gemini 3.5 as one of two newly released models highlighted at the event. (blog.google) Sundar Pichai’s keynote recap linked Gemini 3.5 Flash to Google’s wider product strategy, saying Search would use the model with Google Antigravity to build custom experiences for users’ questions. Google also said the Gemini app would become more agentic, with 3.5 Flash positioned as the next-generation model behind those changes. (blog.google) ### Why is Google emphasizing “agentic” and coding workflows? Google DeepMind’s developer blog said Gemini 3.5 Flash is meant to support “real-world agentic workflows,” describing the company’s focus as a shift “from prompts to action.” The post said the model is intended for tasks where systems need to reason, use tools and complete multi-step work quickly enough to be practical in products and developer environments. (blog.google) Google’s public materials repeatedly paired the model with coding. The company said 3.5 Flash “outperforms Gemini 3.1 Pro across almost all benchmarks,” and framed those gains around coding and agentic performance rather than general chatbot use. Outside coverage from MarkTechPost and other outlets echoed that positioning, describing the launch as a model tuned for AI agents and software development workloads. (blog.google) ### What speed and benchmark claims did Google make? Google said Gemini 3.5 Flash runs “four times faster than other frontier models,” a claim repeated in its I/O developer highlights. In the same post, Google said the model outperforms Gemini 3.1 Pro on almost all benchmarks, presenting speed as a central selling point for high-frequency and latency-sensitive tasks. (blog.google) MarkTechPost reported that Google pitched the model as faster and cheaper, and said it outperformed Gemini 3.1 Pro on coding and agentic benchmarks. The outlet cited benchmark figures including 76.2% on Terminal-Bench 2.1, 1656 Elo on GDPval-AA and 83.6% on MCP Atlas. Those figures appeared in secondary reporting; Google’s official I/O posts available in search results emphasized comparative performance and speed but did not surface the full benchmark table in the snippets reviewed. (blog.google) ### How does this fit into the rest of Google I/O? Google’s May 19 keynote was not limited to developer models. The company’s I/O news hub said the event also included Gemini Omni, updates to Search and developer tooling, and broader product announcements. WIRED’s event recap said Google also unveiled smart glasses expected this fall, underscoring that Gemini was presented as infrastructure across consumer and developer products rather than as a standalone chatbot update. (marktechpost.com) The Gemini app update made the same point in product terms. Google said more than 900 million people now use Gemini monthly across 230 countries and more than 70 languages, and it presented 3.5 Flash as part of the next phase of that rollout. ### Where will developers and users see Gemini 3.5 Flash next? Google said Gemini 3.5 Flash is being released across its Gemini ecosystem, including developer-facing tools highlighted during I/O. (blog.google) The company’s I/O collection page and developer highlights pointed readers to Google AI products and Antigravity-related tooling as the launch continues. (blog.google) Google’s keynote recap said Gemini 3.5 Flash will also power parts of Search and other Google experiences announced on May 19. That makes the next phase of the rollout less about a single demo and more about where Google inserts the model across products developers and consumers already use. (blog.google 1) (blog.google 2)