The majority of Google’s technical advancements are announced at the I/O event every year. But this time, the tech giant didn’t wait until the event to unveil its latest advancements. The company has released an update to its most powerful coding model, and the Gemini 2.5 Pro (I/O edition) is now live for early access. It is designed for developers and claims to improve Gemini’s performance across coding tasks, code editing, and even complex agent workflows.
We just dropped Gemini 2.5 Pro (I/O edition). It’s our most intelligent model that’s even better at coding.
— Google Gemini App (@GeminiApp) May 6, 2025
Now, you can build interactive web apps in Canvas with fewer prompts.
Head to https://t.co/BCZTezi8r9 and select “Canvas” in the prompt bar to try it out, and let us know… pic.twitter.com/gjDpf5nkRE
The new model now ranks #1 on the WebDev Arena Leaderboard, surpassing the 1400 Elo barrier, and also leads the LMArena for coding. It’s available via Gemini API in Google AI Studio and Vertex AI and powers new features inside the Gemini app, like Canvas.
But what’s more exciting than the model itself? What developers are already doing with it.
Here are four apps that showcase just how powerful the new Gemini 2.5 Pro really is:
1. Dictation App
Turn your rambling thoughts into clean messages with the new vibecoded Dictation app, built using the updated model. Enjoy details like wavelength animations and responsive design while maintaining steerability. pic.twitter.com/HenRNw2W0m
— Google AI Developers (@googleaidevs) May 6, 2025
This app was built by Google’s own AI team. It helps users speak in messy and rambling thoughts. After that, Gemini 2.5 Pro cleans them up and puts them into polished, clear messages. However, unlike regular speech-to-text tools, it includes responsive design, wavelength animations, and steerability that allows users to guide tone and style. The company also shared a glimpse of how Gemini can merge multimodality with frontend design.
2. Video-to-Learning App
Google’s AI team also showcased this app, which turns video content into educational experiences like quizzes, flashcards, and summaries. The tool was built entirely inside Google AI Studio, which shows Gemini’s amazing video understanding capabilities (scoring 84.8% on VideoMME).
3. Gemini 95 Media Player
Google’s AI team also showcased a demo, injecting a fully styled media player into the Gemini 95 starter app and making it match the original design perfectly. It’s impressive because there is no manual CSS tweaking or broken layouts. It’s like having a design-aware junior dev on your team.
4. Sidewalk Simulation
I made a real time sidewalk simulation with real time shadows, procedurally generated names, occupation and even a real time heart rate for the characters so every character is different, its been more than incredible. I have some more work im doing so stay tuned, Thanks Gemini pic.twitter.com/q1RXK0URAm
— Marco Kelly (@MarcoKelly_23) May 6, 2025
However, not everything was showcased by Google’s AI team only. An Indie developer shared on X how he created a virtual sidewalk scene with real-time shadows, procedurally generated names and jobs, and even a real-time heart rate for each character.
The interesting part is that every character behaves uniquely, and the dev says Gemini helped bring it to life effortlessly. “It’s been more than incredible,” said. And this is still early days.
Final Thoughts
So, with the newly arrived Gemini 2.5 Pro (I/O edition), Google is giving developers the tools to build faster and smarter by combining coding with design-aware outputs, real-time interactivity, and native multimodality. And the dev community is clearly ready. If this is what was possible before Google I/O, we can only imagine what’s coming next.
