Google will take the stage in California on Tuesday to kick off this year’s Google I/O event. The main keynote will almost certainly focus on Gemini AI advancements, similar to what Google did last year.
After all, Google just gave the world the big Android announcements that would have been the focus of any pre-AI-era I/O event, a sign that there’s no time to focus on the underlying operating system when the Gemini AI features are more pressing.
With Google’s event just around the corner, I can’t help but wonder whether OpenAI will drop an exciting last-minute ChatGPT announcement to steal the spotlight from Google’s I/O 2025 event like it did last year.
The ChatGPT surprise before I/O
OpenAI hosted a ChatGPT livestream a day before Google I/O 2024 kicked off, unveiling the GPT-4o multimodal model that brought support for file uploads in prompts, image understanding, and the incredibly exciting Advanced Voice Mode that lets you chat with the AI just like you talk to people.
This wasn’t a developer event, and OpenAI didn’t have an audience. But just like that, OpenAI shocked the world by beating Google to the punch with a few key advancements.
A day later, Google unveiled the multimodal Gemini 1.5 Pro model with a massive context window (1 million tokens) and demoed Project Astra, the equivalent of Advanced Voice Mode, which would later become Gemini Live. Project Astra also lets the user give “eyes” to the AI via the phone’s camera or head-worn smart glasses.
Fast-forward to Google I/O 2025, and there’s no sign that OpenAI might host a special event on Monday, a day before Google’s Gemini-focused keynote begins.
Where is GPT-5?
One might argue that OpenAI made various big announcements in the past week. GPT-4.1 is now available inside ChatGPT, a model OpenAI recently launched for developers. The AI was only available through the API. On Friday, OpenAI made another announcement targeting developers: the new Codex AI agent.
These are certainly big developments from OpenAI, but they don’t quite qualify as shock-and-awe announcements. That’s what the GPT-4o and Advanced Voice Mode reveal was ahead of last year’s I/O event. Those features targeted most ChatGPT users, not just developers.
Then again, making GPT-4.1 available in ChatGPT and bringing out the new Codex AI agent ahead of I/O might be the kind of moves OpenAI needs to convince developers to use its AI tools instead of whatever Google announces this week.
Whatever Google is about to unveil, it has to be big. After all, Google didn’t even wait for I/O 2025 to roll out the Gemini 2.5 Pro update a few days ago, which also targets coders. That likely means it has even bigger plans for the upcoming show.
So, if a surprise event were in the cards, what could OpenAI even announce on Monday? Well, the company has launched a few ChatGPT models that are currently in testing. GPT-4.5 will replace GPT-4o, and it’s currently in preview. Then we have the new ChatGPT o3 and o4-mini reasoning models. The former is my go-to AI for most of my ChatGPT chats.
OpenAI could make these models more widely available. It could take GPT-4.5 out of preview mode and announce a few new features.
Also, OpenAI could give us more details about GPT-5, the next big ChatGPT upgrade, which is set to launch this summer. I’ll point out that the GPT-5 release has been delayed by a few months, and we still don’t know what this AI might offer.
What could stun us at this point? Well, how about a ChatGPT model that knows what tool to use and when to use it? An AI that decides when to reason, when to search the web, and when to use Canvas to answer the user’s prompt. Maybe it’ll choose on its own to create a Deep Research report or use other agents.
We know OpenAI wants to offer that kind of functionality, but we don’t know when it will be available.
Regarding AI agents, the Operator is still only available to ChatGPT Pro users. That’s the AI agent that can browse the web and even make purchases for you. Launching a version of Operator for the ChatGPT Free and Plus tiers could be the kind of move that plays well ahead of I/O 2025.
Then again, OpenAI doesn’t have to repeat last year’s stunt. ChatGPT remains the main chatbot in town, and it’ll probably stay that way for a while.
Google is playing the same game
If anything, Google proved in December that it wasn’t comfortable with OpenAI’s ChatGPT announcements. Google unveiled Gemini 2.0, its first AI agents, and the Android XR platform, while OpenAI was dropping a new ChatGPT announcement every day before Christmas.
In hindsight, some of those OpenAI reveals were major, so it makes sense for Google to bring Gemini news to the forefront. Google’s AI agents and Android XR clearly weren’t ready for prime time. Nearly six months later, those products are still unavailable to customers, and we might learn more about them at Google I/O 2025.
Then again, we’ve seen plenty of reports suggesting that big AI firms like OpenAI and Google are struggling to deliver the next big wave of AI innovations. That could explain the GPT-5 delays and the absence of Android XR hardware in stores, from Samsung or other brands.
Maybe the real reason OpenAI won’t host a surprise ChatGPT event on Monday is that its development schedule just doesn’t allow for product reveals done purely for marketing effect.
