Google 展示开发者用 Gemini 3.8 Flash 做的四个项目
See what 4 builders are making with Gemini 3.8 Flash
Google 介绍 Gemini 3.8 Flash 自本月发布以来的社区实践,称其在软件工程、智能体任务和专门领域的多步推理上较 3.7 Flash 有明显提升,接近更高成本的前沿模型。
官方给出 3.8 Flash 在软件工程与多步推理上的提升,并用四个开发者案例展示实际用法。
Sep 28, 2026
|
Gemini 3.8 Flash delivers substantial gains in advanced reasoning at scale with Flash-level latency. See what developers have been building and experimenting with it.
Lindsey Lanquist
Contributor
Listen to article
[[duration]] minutes
This content is generated by Google AI. Generative AI is experimental
Gemini 3.8 Flash is our most intelligent workhorse model, delivering significant improvements from 3.7 Flash across software engineering, agentic tasks, and critical, multistep reasoning in specialized domains. These performance gains stem from a core design choice: 3.8 Flash works harder. On complex tasks, it exhibits greater diligence — executing extra reasoning steps and calling tools iteratively to land a more accurate output.
Builders have been creating with Gemini 3.8 Flash since we launched it and new audio models this month. From developing simulations to testing visual reasoning, check out some of the coolest builds and experiments from the community so far.
Create a model rocket simulation.
Ashutosh Shrivastava (@ai_for_success on X) used Gemini 3.8 Flash and Google Antigravity to map the live paths of satellites, space stations, and orbital rockets.
Animate an ink painting.
Noctus (@noctus91 on X) used Gemini 3.8 Flash’s multimodal capabilities to transform Seigaiha waves — a traditional Japanese pattern — into what looks like a moving ink painting.
Create a 3D dinosaur skeleton.
Gemini 3.8 Flash delivers substantial gains on multistep reasoning in STEM and other fields. Emily (@IamEmily2050 on X) pushed these capabilities, asking the model to create a T. rex skeleton with a four-phase prompt, complete with strict requirements, banned shortcuts, and accuracy checks.
Build an automatic transmission simulation from scratch.
Hakm (@hakmgpt on X) gave Gemini 3.8 Flash an engineering challenge: Build an interactive model of an automatic transmission. His prototype involves 10 camera views, four display modes, smart labels, and a side panel that teaches you more about any transmission part you click on.
Gemini 3.8 Flash delivers substantial gains from 3.7 Flash, often approaching the performance of higher-cost frontier models. Try it across Google products and tools, including Google Antigravity and Google AI Studio.
Get the latest news from Google in your inbox
Sign up for our newsletters with product updates, event information, special offers, and more.
Your information will be used in accordance with Google's privacy policy. You may opt out at any time.
来源:Google Gemini Blog · blog.google