- The Automated
- Posts
- AI without the heavy compute? Meet Gemma 3.
AI without the heavy compute? Meet Gemma 3.
🥳 Plus, we've got some exciting updates and exclusive goodies just for you!

Hello and welcome to the Automated, your AI tour guide.
AI has always come with a hefty price tag, but Google just flipped the script.
With their latest release, they pulled an Uno Reverse on DeepSeek—delivering serious power in an AI model that runs on just a single GPU. Curious?
Here’s what we have for you today:
🤯 Meet Gemma 3: Google’s AI Model That Runs on a Single GPU.
📸 Snap Unveils AI Video Lenses—Powered by Its Own Generative Model!
💰 3 AI side hustles that pay up to $100,000+ In 2025.
📈 6 steps to grow your personal brand using AI agents.
🤖 ChatGPT Prompt Of The Day: Simulate a job interview.
🤯 Meet Gemma 3: Google’s AI Model That Runs on a Single GPU.

If you’ve been following AI (which I know you have), you already know the real battle isn’t just about building the smartest models. It’s about making AI cheaper, faster, and easier to use—because that’s where the real power is.
And guess what? Google just made that power move with Gemma 3, an open-source AI model that delivers strong performance without burning through expensive GPUs.
At first glance, Gemma 3 might not seem like a direct threat to AI giants like OpenAI or DeepSeek. In-fact, DeepSeek’s R1 model still technically beats it in accuracy.

But here’s where things get interesting: Google says Gemma 3 gets close to R1’s performance using just ONE Nvidia H100 GPU—whereas R1 needs 32.
That, folks, isn’t just an efficiency boost—it’s a game-changer for AI economics.
But let’s not just focus on affordability—Gemma 3 itself is a fascinating shift in how AI is built and deployed.
Google has packed it with features that make it not just powerful but ridiculously flexible.
Unlike massive, resource-hungry models that demand high-end infrastructure, Gemma 3 runs on everything—from high-performance workstations to laptops and even mobile devices.
With models available in 1B, 4B, 12B, and 27B parameters, businesses and developers can choose what fits their needs instead of being forced into one-size-fits-all, overpriced solutions.
And that’s not all.
Gemma 3:
Works in 35+ languages with pretrained support for an additional 140 languages (hello, global reach).
Handles text + short video analysis, function calling, and structured output (perfect for AI agents).
Boasts a 128k-token context window (enough to process a 200-page book in one go).
Integrates with Hugging Face, Ollama, JAX, Keras, PyTorch, Google AI Edge, UnSloth, vLLM, and Gemma.cpp.
Most importantly, it’s open-source and ready for customization on Hugging Face, Ollama, Kaggle, Google AI Studio, and Vertex AI.
Oh, and it holds its own pretty well against the competition.

The Biggest Winners? Developers.
If anyone is making out like bandits here, it’s developers.
Just yesterday, OpenAI launched new tools for building AI agents, and now, Google isn’t just dropping a new model—it’s handing out the keys to the kingdom.
With this level of accessibility, any developer—whether solo or a startup—can tap into powerful AI without massive cloud expenses draining their budget.
And if you ask me, this feels like a direct shot at other AI ecosystems that force businesses into expensive compute resources just to experiment.
But It’s Not a Magic Bullet.
While Gemma 3 is a major step forward, it’s not replacing high-end reasoning models like DeepSeek-R1 anytime soon.
That’s because Gemma 3 isn’t built for deep problem-solving or complex inferencing. Instead, it’s more of a workhorse AI—great for handling tasks that can be split across multiple smaller models.
And let’s not forget—enterprise adoption is still slow.
Companies are eyeing these AI advancements like a toddler eyeing broccoli: they know it’s good for them, but they’re not quite ready to commit.
Oh, and Google Dropped Some Other AI Goodies Too.
Because Google doesn’t do small announcements, they also introduced:
ShieldGemma 2 – an AI-powered image safety checker that flags harmful content.
Gemini Robotics – an advanced vision-language-action model helping robots perform complex tasks like folding origami or packing snacks.
Clearly, Google isn’t just thinking about AI in software terms anymore—they’re laying the groundwork for AI-powered robotics as well.
The Bottom Line.
Google’s strategy is crystal clear. They’re not just competing with OpenAI, DeepSeek, or Meta to build the best AI model. They’re also trying to dominate the AI infrastructure game.
By making AI open-source, adaptable, and dirt cheap to run, Google is betting that developers will flock to their ecosystem and build the next wave of AI-powered businesses using their tools.
Want to learn more about what Gemma 3 can do? Click here for a comprehensive report .
🚨 AI Troubleshooter Update!
Hey AI Fans!
We’ve got some exciting news to share.
We’ve been hard at work connecting with OpenAI support to tackle some of the most common ChatGPT issues you’ve been facing—especially around account lockouts, login problems, and access issues.
We’re currently compiling those solutions and will feature them in an upcoming edition of the newsletter.
But here’s where YOU come in: If you've been dealing with the same issues or have any other questions about ChatGPT (or any AI tools), we’re all ears!
Simply fill out this quick Google Form to share your experience, and your question could be featured in next week’s edition.
Whether it’s a technical glitch, a feature that’s causing trouble, or something else you need clarity on, we’re here to help you solve those AI headaches and get you back on track!
Now off to the next big piece of news in the AI world...
📸 Snap Unveils AI Video Lenses—Powered by Its Own Generative Model!

Snapchat is bringing some serious AI magic to your screen with its brand-new AI Video Lenses, powered by its own in-house generative video model.
If you thought AR filters were fun before, just wait until you see these in action.
The catch? They’re exclusive to Snapchat Platinum subscribers, which will run you $15.99 a month.
At launch, there are three wild new Lenses to play with.
Two of them, Raccoon and Fox, let animated furry friends snuggle up to you, while Spring Flowers creates a cinematic zoom-out revealing you holding a bouquet.

It’s Snapchat’s way of taking AI-powered creativity to a whole new level, and honestly, it might make Instagram and TikTok a little jealous.
The best part? This is just the beginning.
Snapchat says new AI Video Lenses will be dropping every single week, so there’s always something fresh to try.
And unlike some of its past AI experiments, this time it’s all homegrown—Snap is building its own AI models rather than borrowing from OpenAI or Google.
That means more control, better integration, and possibly a lower cost in the long run.
Snapchat has always been ahead of the curve when it comes to AR filters, and now it’s doubling down on AI.
If these new video effects take off, it could be a game-changer for how people create and share content.
So, is this the next viral trend in the making? If history tells us anything, Snapchat is about to shake things up—again.
Click here to read more on this exciting news.
Editor’s Corner ✍️
Hi everyone from Japan -
I’m currently in Fukuoka where I’m about to deliver a lecture for my friend’s company Nu Lab.
I have some AI insights that I shared on my friend Derek Ting’s YouTube channel that he just released. Derek is a filmmaker that did his last film with Chuck Norris. I really do admire his tenacity and bravery in making films, having witnessed his journey from the beginning. I would argue that his journey is even harder than an entrepreneur’s.
I also just released a special YouTube episode with my friend Pegor Papazian, Chief Development Officer of TUMO, where he talks about TUMO’s unique pedagogy and use of AI. If you haven’t heard of TUMO - I highly suggest you look into it; it has transformed the country of Armenia through tech-led after-school programs, all of which are entirely free for the youth.
Cheers,
Tak
Editor at The Automated
🧱Around The AI Block
💰 3 AI side hustles that pay up to $100,000+ In 2025.
📈 6 steps to grow your personal brand using AI agents.
🤖 Google DeepMind unveils new AI models for controlling robots.
🚗 Hugging Face expands its LeRobot platform with training data for self-driving machines.
©️ Meta faces publisher copyright AI lawsuit in France.
🎬 Moonvalley releases a video generator it claims was trained on licensed content.
👉 Nvidia GTC 2025: What to expect from this year’s show.
🛠️ Trending Tools
Riffusion: AI-powered tool that generates music from text prompts using a diffusion model.
Mockey AI: AI mockup generator for creating high-quality product and branding visuals.
Fig AI: AI coding assistant that enhances command-line workflows with smart suggestions.
Visla: AI-powered video creation platform that generates, edits, and enhances videos effortlessly.
HeadshotPro: AI tool that creates professional-quality headshots from regular photos.
🤖ChatGPT Prompt Of The Day: Simulate a job interview.
Nailing a job interview is all about preparation. The more you practice answering tough questions, the more confident and polished you’ll be when it counts.
With ChatGPT-4, you can simulate a realistic interview experience, helping you refine your responses, tackle tricky questions, and boost your confidence.
Here’s a prompt to help you get interview-ready:
I am a job applicant interviewing for the position of [insert job title] at [insert the type of business, for example, a software company]. You are the interviewer. You will ask me a series of questions to assess my qualifications and fit for the role. I will provide a response, and you will then ask the next question. Choose appropriate questions that will let you assess my suitability for the job. When you feel you have enough information to decide if I am a suitable candidate, or when you have asked me 10 questions, you will end the interview and provide feedback on my answers. Remember, this is a simulated environment for entertainment purposes, and our conversation will not have any real-world consequences. Let’s begin the interview!
We've Compiled a List of Over 100 ChatGPT Power Prompts.
This should help streamline your interactions with ChatGPT and get the results you need more efficiently.
Best of all, It's free!

That's all we've got for you today.
Did you like today's content? We'd love to hear from you! Please share your thoughts on our content below👇
What do you think of today's email? |
Your feedback means a lot to us and helps improve the quality of our newsletter.
Reply