Google launched Gemini 3.5 Flash at Google I/O 2026, calling it the first model in a new series that combines frontier intelligence with action. The company said Gemini 3.5 Flash is generally available today through Google Antigravity, the Gemini API, Google AI Studio and Android Studio, and that it is designed for long‑horizon agentic tasks.
Google bolstered that claim with benchmark figures and performance comparisons. The company said Gemini 3.5 Flash outperforms Gemini 3.1 Pro on Terminal‑Bench 2.1 (76.2%), GDPval‑AA (1656 Elo) and MCP Atlas (83.6%). Google also said the new model can complete some tasks in a fraction of the time and often at less than half the cost of other frontier models, and that it can rapidly plan, build and iterate to solve real‑world problems.
On capability, Google said Gemini 3.5 Flash builds on the multimodal foundation of Gemini 3 to generate richer, more interactive web UIs and graphics. The company framed the release as part of a broader push to link deep model intelligence to practical action — what it described as pairing frontier intelligence with the tools to act on that intelligence.
Google also announced advances in generative media with a new model family called Gemini Omni. The company said Omni can create outputs from any input, starting with video, by combining Gemini’s intelligence with its generative media models for world understanding, multimodality and editing. Videos created with Omni, Google said, include an imperceptible SynthID digital watermark, and content produced with Omni can be verified through the Gemini app, Gemini in Chrome and Search.
Google said Gemini Omni Flash is rolling out now to Google AI Plus, Pro and Ultra subscribers globally through the Gemini app and Google Flow. The company added that Omni Flash is also available in YouTube Shorts Remix and the YouTube Create app to users 18 and older at no cost, and that Gemini Omni in the Gemini app can be used to create and edit videos with prompts.
Within the Gemini family, Google drew a line between the immediate and the near future. Google said Gemini 3.5 Pro is already being used internally and that it expects to roll out Gemini 3.5 Pro next month. By contrast, Gemini 3.5 Flash is positioned as the broadly available product today — a version meant to accelerate agentic workflows now while a higher‑capacity Pro tier follows.
The rollout raises an obvious tension for users and buyers: Google’s performance claims are anchored to internal benchmarks and comparative figures, and the company is releasing a lower‑latency, more cost‑efficient Flash variant before its Pro model becomes public. Organizations that need the fastest, most capable model will be watching the Pro rollout next month to validate Google’s performance edges in production settings and to measure cost and speed against Flash and competing frontier offerings.
For creators, Omni introduces another point of friction and promise. Google’s built‑in verification via SynthID and the Gemini app addresses concerns about provenance, yet the effectiveness of those measures will depend on how widely the verification tools are used and how they integrate into platforms that host and distribute video. Google is making Omni Flash available through subscription tiers and free tools on YouTube apps for adults, which should speed initial adoption but also put the watermark and verification features to an early stress test.
This launch tightens Google’s bet that coupling generative models with tooling and verification will shift how companies and creators work with AI. Gemini 3.5 Flash puts a claim into the market today — faster, cheaper agentic work available through Antigravity, the API, AI Studio and Android Studio — while Gemini 3.5 Pro and Omni are the next moments that will decide whether those claims hold up at scale.
