|
Google held I/O yesterday in Mountain View. The keynote ran for two hours, covered everything from Android to XR glasses to Search, and managed to announce more in two hours than most companies launch in a year. The general tech press is writing about Gemini Spark and AI Ultra pricing. For creators working in AI image, video, and music, the story is somewhere else. Five announcements meaningfully change how the work gets done. One of them matters more than the others. Here is what to focus on.
I watched the keynote live and have been testing what is already available since this morning. The summary below is what an AI image and video creator actually needs to know, with the practical takeaways for each piece. The shifts are real, but most of them will not land in your workflow until the rollouts complete over the next few weeks.
|
|
Announcement One
Gemini 3.5 is now the flagship model.
|
|
Google unveiled the Gemini 3.5 lineup as the new flagship family. Gemini 3.5 Flash is available today, and Gemini 3.5 Pro will follow over the coming weeks. The reasoning capabilities improved across the board, the context window expanded, and the model handles multimodal input more cleanly than 3.0 did.
For creator work, the practical change is in how Gemini handles complex prompt instructions. The model now follows multi-step prompting more reliably, which matters most for the chat-based image and video editing workflows that have been emerging across the AI tools landscape. If you have been using Gemini for prompt engineering or for image generation through Nano Banana 2, the upgrade lands automatically. Your existing prompts should produce slightly cleaner output, and instructions that used to need three tries should land in one.
|
|
Announcement Two
Gemini Omni is now a model family.
|
|
Omni went from a single rumored video model to an entire series of multimodal models. Gemini Omni Flash is available today and accepts image, audio, and video input. Omni combines reasoning with creation in a single architecture, meaning you can chat with the model about an image or video while it generates and edits in the same conversation.
This is the announcement most directly relevant to image and video creators. The chat-based editing pattern that worked in the early Omni leak is now official. Upload a photo, ask for animation, ask for refinements, ask for sound. Each step layers onto the last one. For creators whose workflow has been image-then-export-then-edit-elsewhere, Omni collapses three tools into one conversation. The raw generation quality still trails Seedance 2.0, but the editing fluidity is now the category leader.
|
|
Announcement Three
Gemini Spark is the bigger story.
|
|
Gemini Spark is a 24/7 personal AI agent that lives in the Gemini app and acts across your Google products. Beta opens next week for AI Ultra subscribers in the US. The demos shown on stage included Spark setting up an RSVP tracker in Google Sheets, populating it from emails, then updating it in real time as responses came in. The agent runs in the background and takes complex tasks off your plate without requiring step-by-step prompting.
This is the announcement that matters more than the rest for creators. The shift from prompting tools to delegating to agents changes what creator work looks like in twelve months. Imagine asking Spark to manage your social content calendar, generate weekly image variations through Omni, post them on schedule, and track which ones drove the most engagement. Not as a separate operation. As one delegation. The infrastructure is finally arriving for the kind of small-team-of-one creator practice that has been theoretically possible for two years and is now becoming practical.
|
|
Announcement Four
The new pricing model rewards heavy use.
|
|
Google moved the Gemini app from daily prompt limits to a compute-based pricing model. Limits refresh every five hours until you reach a weekly cap. Simple text prompts use less compute than a complex video generation, which means heavy creator workflows are no longer punished by hitting arbitrary prompt counts. The AI Ultra tier was also restructured: a new $100 per month plan gives 5x the usage of Pro, and the existing Ultra is now $200 with the same capabilities as before.
For creators making video and complex image work, the math is more favorable than it sounds. The previous prompt-count model penalized creators who iterate, which is exactly what good AI image work looks like. The compute model lets you generate twenty variations of a clip on Monday and a few simple chat prompts on Tuesday without burning through your allotment unfairly. If you have been on AI Pro and felt constrained, the new Ultra at $100 is probably the right tier. If you are at the Pro level and not constrained, stay where you are.
|
|
Announcement Five
Daily Brief and information agents.
|
|
Daily Brief rolled out today across Plus, Pro, and Ultra in the US. It pulls from your Gmail, Calendar, and Tasks to produce a personalized morning digest that prioritizes what needs attention and suggests next steps. The first version is rough, but the direction is clear. Information agents are arriving across Search and Gemini this summer to run continuous background tasks for users.
For creators, these are the features that quietly add up over time. Daily Brief is not a creator tool. It is a productivity tool that gives you back an hour a day, which is the hour you actually use for creative work. The information agents in Search will do the same job for research. A summer of small productivity gains adds up to a meaningfully different operating model by the time the holiday season hits.
|
|
What to Do This Week
Three concrete moves before the rollouts complete.
|
|
One. Spend an hour with Gemini Omni Flash. Animate three photos. Try chat-based editing on one of the outputs. The point is to feel where the conversational editing workflow lands compared to the prompt-first workflow you have been using. The shift is real and worth experiencing before you decide how to route work between Omni and your existing tools.
Two. If you do not already have an AI Ultra plan and you generate AI video regularly, look at the new $100 tier. The compute-based pricing is more favorable for creator workflows than the previous prompt limits, and Spark beta access alone is worth experimenting with. If you are happy on Pro, stay there for now.
Three. Get on the Gemini Spark beta list when it opens next week. The agentic shift is the biggest thing announced at I/O, and being early gives you twelve months of head start on creators who wait for the general release. Even if Spark turns out to be more limited than the demos suggested, the muscle of delegating to an agent rather than prompting a tool is going to matter.
|
|
The deeper pattern across all five announcements is that Google is reshaping what it means to work with an AI tool. Less prompting. More conversation. Less single-shot generation. More iterative refinement. Less you-doing-the-work-and-the-AI-helping. More you-directing-the-AI-and-checking-the-output. The category is moving from tools to collaborators, and Spark is the clearest signal yet that the shift is happening this year rather than next.
For creators who have been building careers around AI image and video work, this is the moment to start practicing the agentic workflow even before the tools fully arrive. The skills that will matter most twelve months from now are not just prompt craft. They are task design. They are knowing what to delegate and what to keep in your own hands. Start practicing now while the stakes are low.
|
|
•••
I am writing a deeper piece on how to set up creator workflows that take advantage of Gemini Spark when it goes broadly available. Specific delegation patterns. Real task examples. What to keep manual.
Want it when it ships? Reply with send me the Spark playbook and I will get it to you.
|
|
A QUESTION FOR YOU
Which of the five matters most for your work?
Reply and tell me. The replies determine which announcement I dig into deeper next, and which workflow I cover first.
If this resonated, forward it to a creator trying to keep up with the I/O announcements.
|
|
Until next time,
Luxe Prompting
|
|
Luxe Prompting
AI Image Generation for Creators
|
