This NEW Google Gemini Update Is INSANE!

Google just dropped something absolutely crazy with Gemini. They increased file size limits from 20 megabyt to 100 megabyte and they added support for Google Cloud Storage and direct URLs. This means you can now handle massive files without the upload nightmare. If you're building AI apps, this changes everything. Let's dive in. All right, let's talk about this Gemini update because it's genuinely insane. Google rolled this out around January 8th, 2026. And honestly, it's one of those updates that makes you go, "Finally. " So, here's what changed. Before this update, if you wanted to use Gemini with large files, you were stuck with the files API, and the files API had limits. 20 megabytes for inline files, files disappeared after 48 hours. If you were building production apps with videos or big documents, you had to keep re-uploading everything. It was a pain.

Now, Google fixed all of that. Three massive changes. First, inline file size went from 20 m to 100 mAB. That's five times bigger. You can now drop larger images, short video clips, audio files directly into your API calls without any pre-upload. Second, they added support for external URLs. This is huge. You can now pass public links or pre-signed private URLs directly into Gemini. So if you've got files sitting in AWS S3 or Azure, you just generate a signed URL and pass it straight to Gemini. No downloading to your back end, no storing it somewhere else. Gemini fetches it securely for you. Third, Google Cloud Storage integration. You can now register files from your GCS buckets. Once you register them, you get a persistent file URI that you can reuse across multiple requests. No more uploading the same file over and over. This is perfect for production apps where you're processing the same videos or documents repeatedly. Let me give you

a real example. Let's say you're building an AI automation tool for the AI profit boardroom. You want to analyze customer testimonial videos to extract key pain points and success stories. Before this update, you'd upload a 50 m video to the files API. It would expire in 48 hours. If you wanted to run different prompts on that same video, you'd have to re-upload it. Now you upload it once to Google Cloud Storage, register it with Gemini, and you can analyze it unlimited times without touching the file again. You could run one prompt to extract pain points, another prompt to identify success metrics, another to generate social media clips, all from the same registered file. That's the power of this update. And it works with multimodal content, videos, audio files, PDFs, images, all of it. Now, I want to pause here for a second because this ties directly into what we teach in the AI profit boardroom. If you're watching this and thinking, I need to learn how to actually use these AI tools to save time and automate my business, then you need to check out the AI profit boardroom. We show you exactly how to use tools like Gemini to build real automation workflows, not just theory. Actual step-by-step processes you can implement today. Whether you're creating content, analyzing data, or building AI powered tools, we've got you covered. Link in the description. Now, let me

show you how this actually works in code. Don't worry, I'll keep it simple. Let's start with external URLs. Say you've got a landing page design PDF stored in AWSS3. You generate a pre-signed URL using Boto 3 or whatever tool you use. Then in your Python code, you do this. You import the Gemini SDK. You create a part object using part from URI and pass your URL. Then you combine that with a text prompt, something like analyze this landing page design for the AI profit boardroom and suggest improvements to increase conversions. You call generate the content with both the file part and the prompt. Gemini fetches the file from your URL and processes it. No backend downloads, no temporary storage. It just works now for

Google Cloud Storage. It's even better. First, you authenticate using Google. default. Make sure you've got the right oorthth scopes like dev storage readon. Then you register your file. You call client. files register alert files and pass your GCS URI. Something like GS my bucket or testimonial video MP4. Gemini registers it and gives you back a file object with a persistent URI. Now you can use that URI in any generation request. You could analyze the video today, run another prompt tomorrow, use it in a batch job next week. The file stays registered and ready. This is perfect for building AI automation workflows. Let's say you're creating content for the AI profit boardroom YouTube channel. You've got raw footage in GCS. You register it once. Then you run multiple prompts, generate video chapters, extract key quotes, create short form clips, summarize the main points, all from the same registered file. No re-uploads, no wasted time, just pure automation. And here's something else that's cool. The inline file limit increase to 100 m. This is great for quick prototyping. Let's say you're testing a new AI tool for analyzing competitor websites. You take a full page screenshot. Maybe it's 80 ml by because it's high resolution. Before you couldn't send that inline, now you can basease 64 encode it and drop it straight into your API call. Perfect for testing and iterating fast. Let's talk

about why this matters for production apps. Before this update, the files API was really designed for prototyping. You could upload a file, test your prompts, see what works. But for production, it wasn't ideal. Files expired. You had to manage uploads constantly. It didn't scale well. Now with GCS integration and external URLs, you can build production scale apps. Imagine you're running an AI automation agency. You've got clients sending you videos, PDFs, audio recordings every day. You store everything in Google Cloud Storage. You register the files once. Then you run your AI workflows automatically. No manual uploads. No expiration timers. Just clean scalable automation. And the external URL support means you can work with files from anywhere. AWSS3, Azure, blob storage, your own CDN. As long as you can generate a URL, Gemini can process it. This opens up so many possibilities. You could build a tool that monitors your company's shared drive. Whenever a new document gets uploaded, it automatically gets analyzed by Gemini. Extract key points, generate summaries, flag important information, all automatically. Or let's say you run the AI profit boardroom community. Members upload their business documents. You process them with Gemini to give personalized feedback, suggest improvements, identify growth opportunities, all powered by this new GCS integration. Now, a few technical notes for GCS. You need to set up IM permissions correctly. Your service account needs read access to the bucket. The official docs walk through the whole authentication process. You'll need the right OOTH scopes, and you'll want to test everything in development first. Google also has an AI Studio demo uplet where you can test external URLs live just to see how it works before you write any code. I'll drop that link in the description. Uh, make sure you're using the latest version of the Google Genai SDK. They rolled out these features with the newest SDK versions. If you're on an older version, update it. The change log on the official Gemini API docs has all the details on which models support which features. One more thing about inline files, the 100 me limit is great, but remember it's base 64 encoded. So your actual file might be a bit smaller, but still going from 20 m to 100 nmies is massive. It means you can prototype with real world file sizes, not just tiny samples. This makes testing so much faster. You can try different prompts on actual production data, see what works, iterate quickly, then move to GCS for the final workflow. Now, let's talk about use

cases because this update unlocks so many new possibilities. First, video content analysis. You're creating YouTube videos for your business. You upload raw footage to GCS, register it with Gemini, then run prompts to generate chapters, extract quotes, create descriptions, all automated. Second, document processing at scale. You've got thousands of PDFs you need to analyze. Upload them to GCS. Register them all. Then batch process with different prompts, extract data, summarize content, flag important sections. Third, audio transcription and analysis, customer support calls, podcast episodes, webinars, store them in GCS, register with Gemini, then analyze for insights, generate transcripts, identify key moments. Fourth, image analysis for marketing, your split testing landing pages. You've got dozens of design variations. Upload screenshots to GCS, register them, then use Gemini to analyze which designs have better hierarchy, clearer CTAs, stronger messaging. All these use cases were technically possible before, but they were clunky. Now they're smooth and production ready. And here's the thing that gets me most excited. This update makes AI more accessible for smaller teams. Before you needed complex backend systems to handle large files, storage servers, upload managers, expiration tracking. Now you just need a GCS bucket and a few lines of code. That's it. This levels the playing field. Small agencies can build the same powerful automation tools that big companies use. And if you're a solo entrepreneur running an AI automation business, this is perfect. You can offer sophisticated document processing services without hiring a dev team. Just learn the Gemini API, set up GCS, and start building. That's the power of this update. It removes friction, makes things simpler, and lets you focus on building value, not managing infrastructure. So, if you want

to learn the full process, get the SOPs, and access over 100 AI use cases just like this one. Join the AI profit boardroom. We break down exactly how to use tools like Gemini to automate your business and save hours every week. No fluff, just practical workflows you can implement today. link in the description. And if you want to go even deeper, check out the AI Success Lab. It's our free AI community where you can get all the video notes, connect with 40,000 members who are crushing it with AI, and get access to even more resources, links in the comments and description. Right, that's the Google Gemini update, 100 millm of my inline files, GCS integration, and external URL support. This is huge for anyone building AI apps. If you found this helpful, hit that like button and subscribe for more AI updates. And seriously, drop a comment below. Tell me what you're going to build with this. Julian Goldie reads every single comment. I'll see you in the next one.