Automate Basics
New model · Google

Google shows what builders can make with Gemini 3.8 Flash

By Automate Basics, written with AI from Google's original

· 2 min read

Hands arrange metal gears beside a diagram and digital transmission model
Image: AI-generated illustration.

Google says Gemini 3.8 Flash improves on its previous Flash model in software engineering and complex reasoning. The company is showing builder projects and says people can try the model in Google AI Studio.

What changed with Gemini 3.8 Flash

Google says Gemini 3.8 Flash is better than Gemini 3.7 Flash at software engineering, tasks that use tools, and problems requiring several reasoning steps. The company says the model can spend more effort on difficult requests, including taking additional reasoning steps and using tools repeatedly.

Google’s latest announcement shows what builders have made with the model rather than introducing a separate feature. The examples include a satellite-path simulation, an animated ink painting, a dinosaur skeleton and an interactive automatic transmission model. Those projects illustrate the kinds of visual and technical tasks people are trying. They do not establish how well the model will perform on a particular workplace task.

Who can use Gemini 3.8 Flash?

Google says people can try Gemini 3.8 Flash in Google AI Studio and Google Antigravity, as well as other integrated Google tools. The announcement does not specify which plans include the model or what using it costs.

The examples are most relevant to people who create prototypes, visual explanations or technical training materials at work. A team developing an interactive explanation of a mechanical process, for instance, could use the transmission project as a prompt for what to investigate. Google describes that project as a builder’s prototype, not a ready-made workplace template. People handling work information should also check their organization’s rules before putting that information into an AI tool.

How do I try Gemini 3.8 Flash at work?

A useful first test for Gemini 3.8 Flash is a small task with a result you can check yourself. Google’s examples suggest asking it to plan an interactive explanation of a process or reason through the requirements for a visual model. Give it the constraints that matter, including anything the result must not do.

Check whether the output follows each requirement, whether the underlying explanation is accurate and whether interactive elements behave as described. Google highlights a builder who used explicit accuracy checks when requesting a dinosaur skeleton. That is a sensible approach for workplace prototypes too: treat the output as a draft to verify, especially when a technical mistake could mislead colleagues.

Frequently asked questions

Is Gemini 3.8 Flash a new tool or a new model?

Gemini 3.8 Flash is a Google AI model. Google says it improves on Gemini 3.7 Flash in software engineering, tool-using tasks and complex reasoning. The builder projects in the announcement are examples of work made with the model, not separate Google products.

Where can I try Gemini 3.8 Flash, and does it cost anything?

Google says Gemini 3.8 Flash can be tried in Google AI Studio and Google Antigravity, along with other integrated tools. Its announcement does not say which plans include access or state a price, so check the terms for the tool you intend to use.

Tools in this piece

Written with AI from Google's original and published after automatic checks: every figure here appears in the original, and no sentence is copied from it. The picture is AI-generated. The original is the authority.

Source: See what 4 builders are making with Gemini 3.8 Flash, Google, 28 Sept 2026.