Gemini 3.5 Flash, released May 19, 2026, pairs frontier — level reasoning with agentic capabilities to plan, execute and iterate on multi — step real-world workflows and is available now across consumer, developer and enterprise channels.
On May 19, 2026, the company unveiled Gemini 3.5 and launched Gemini 3.5 Flash, a new model family designed to combine frontier — level reasoning with agentic action. The release aims to move intelligent agents from planning to supervised execution of multi — step, real-world workflows. That shift could let users and organizations automate complex tasks that previously required extensive human coordination.
Gemini 3.5 Flash is available immediately to billions of users through the Gemini app and AI Mode in Google Search for the general public; to developers via the agent — first Google Antigravity platform and the Gemini API in Google AI Studio and Android Studio; and to enterprises through the Gemini Enterprise Agent Platform and Gemini Enterprise. The company said a higher — tier 3.5 Pro is already in internal use and is scheduled to roll out next month.
Published benchmarks position 3.5 Flash as matching or surpassing other flagship models on agentic and coding tasks. The model outperforms Gemini 3.1 Pro on Terminal — Bench 2.1 (76.2%), GDPval — AA (1656 Elo) and MCP Atlas (83.6%), and scores 84.2% on CharXiv Reasoning for multimodal understanding. The team also reports that 3.5 Flash generates output tokens per second at roughly four times the rate of other frontier models and places in the top-right quadrant of an internal Artificial Analysis index for combined quality and latency.
The release emphasizes agentic scale: 3.5 Flash is tuned for long-horizon problems and, when paired with the Antigravity harness, can run supervised collaborative subagents to perform multi — step workflows. Demonstrated examples include automatically renaming and categorizing unstructured asset libraries, synthesizing the AlphaZero paper and coding a playable game in six hours using two agents, upgrading legacy codebases into Next.js, and generating new city landscapes through coordinated subagent workflows.
Built on Gemini 3’s multimodal foundation, 3.5 Flash supports richer interactive user interfaces and graphics aimed at builders. Developer — facing demonstrations cover creating interactive animations for a research paper, converting plain text into interactive hardware components on AI Studio, running parallel concepts to produce a full branding package, and generating multiple UX approaches for a checkout flow in roughly 60 seconds.
The company frames the practical impact in terms of time and cost: the balance of speed and performance is intended to let developers and enterprises finish tasks that once took days or weeks in a fraction of the time and, in many cases, at less than half the cost of other frontier models. According to the announcement, enterprises and industry partners are already deploying 3.5 Flash to reduce development and auditing toil, improve code maintenance, and scale supervised agent deployments via Antigravity.
Sources
Replies (0)
No replies in this topic yet.
