At Google I/O 2026, Google shipped two new model families: Gemini 3.5 and Gemini Omni. The press coverage went to the flagship reasoning and the flashy video generation. For a small business, the interesting model is the one that is fast, cheap, and smart enough to run all day.
The New Lineup
| Model | What it is | Best for |
|---|---|---|
| Gemini 3.5 Flash | Frontier intelligence at Flash speed and price. Available now in the Gemini app, Search, and the API. | The model you actually run in production: support, content, automation. |
| Gemini 3.5 Pro | Advanced reasoning for heavier workloads. In testing, launching next month. | Complex analysis where you want maximum reasoning. |
| Gemini 3.5 Lite | Maximum reactivity, minimal overhead. | High-volume, low-complexity tasks at the lowest cost. |
| Gemini Omni | Takes image, voice, video, or text in and generates video out. Shipping variant is Omni Flash. | Short video content. Free in YouTube Shorts; paid in the Gemini app. |
Source: Google I/O 2026 announcements. Flash ships now; Pro follows next month.
When Each Model Ships, Including Gemini 3.5 Pro
What Google said at I/O in May 2026:
- Gemini 3.5 Flash: available at the keynote in the Gemini app, Search, and the API.
- Gemini 3.5 Pro: in testing, with a stated plan to launch it next month, which pointed to June 2026.
- Gemini Omni Flash: live at launch in YouTube Shorts Remix, the YouTube Create app, the Gemini app, and Google Flow, with API access rolling out in the weeks after.
Treat the Pro date as Google's stated plan, not a promise. Flagship launch dates can move, so confirm current availability before you build a workflow around it. Flash is the one you can build on now.
Gemini 3.5 Flash: The One You Will Actually Run
The pattern across every frontier lab in 2026 is the same: the "small fast cheap" tier keeps absorbing capabilities that used to require the flagship. Gemini 3.5 Flash is the clearest example yet. Google says it delivers frontier-level intelligence while running about 4x faster than other frontier models in output speed, and it outperforms the previous Gemini 3.1 Pro on coding, agentic, and multimodal benchmarks.
For a small business, that combination is the whole game. You almost never need the absolute smartest model. You need one that is good enough to handle support replies, draft content, qualify leads, and run automations, fast enough that it does not feel slow, and cheap enough to leave running. Flash is built for exactly that, and it is available today in the Gemini app, in Search, and through the API.
Gemini Omni: Video From Almost Anything
Gemini Omni (the variant shipping now is Omni Flash) takes an image, a voice clip, a video, or text and generates video, grounded in real-world physics so scenes hold together. Google is starting with video output and says any-input-to-any-output generation will follow. If social content is part of how you reach customers, this is the most directly useful announcement of the event: a product photo or a few lines of text can become a short video without a production budget.
Where you can use it matters, and the access split is generous. It is free for anyone 18 or older inside YouTube Shorts Remix and the YouTube Create app. Inside the Gemini app and Google Flow it requires a paid Google AI Plus, Pro, or Ultra subscription, and developer access through the API is rolling out in the coming weeks. Either way, generated video still needs a human eye before it represents your brand in public. Treat it as a first-draft engine, not an autopilot.
The Price Move Nobody Should Miss
Google cut its top AI Ultra tier from $250 to $100 per month. That is not a rounding error. It is the same competitive pressure we wrote about when AI got 60% cheaper in two years, still running. Three labs are now racing on price as hard as on capability, and the small business is the beneficiary.
If You Already Live in Google Workspace
Most small businesses run on Gmail, Docs, Sheets, and Drive. The practical takeaway from I/O is not "switch to Google." It is that the AI is now strongest wherever your business already lives. If you are a Workspace shop, Gemini 3.5 baked into the tools you use every day will likely beat bolting a separate AI product onto the side. If you live in Microsoft 365, the same logic points the other way. Pick by your stack, not by the keynote.
The Bottom Line
The flagship models make the headlines, but the small fast model makes the money for a small business. Gemini 3.5 Flash is the one to test first: frontier-quality output, 4x the speed, and a price you can leave running. The video model is a bonus if you publish social content, and the price cut is a reminder that waiting a quarter usually makes AI cheaper, not more expensive.
Sources
- Google, "100 things we announced at Google I/O 2026." blog.google
- Google, "Google I/O 2026: Sundar Pichai's opening keynote." blog.google
- 9to5Google, "Everything Google announced at I/O 2026: Gemini, Search, Android XR, and more," May 19, 2026 (Flash 4x speed, AI Ultra price cut to $100). 9to5google.com
- Google Developers Blog, "All the news from the Google I/O 2026 Developer keynote." developers.googleblog.com
Not sure which model your business should actually run?
We help small businesses match the right model to the right job, wire it into the tools you already use, and keep the cost sane.