- AI Fire
- Posts
- 🤖 OpenAI Upgraded Our Free Tier
In partnership with
OpenAI gave all free users a massive boost with 2 new models that work together like a mini-team on your phone. Plus, join our live workshop tonight guys!!!
What's on FIRE 🔥
IN PARTNERSHIP WITH HUBSPOT
Want to get the most out of ChatGPT?
ChatGPT is a superpower if you know how to use it correctly.
Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.
Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.
AI INSIGHTS
OpenAI just rolled out GPT-5.4 Mini and Nano to free users, just weeks after launching the more powerful GPT-5.4 model. These are 2 new models:
GPT-5.4 Mini: Faster (2× speed vs previous models)
Strong at coding, reasoning, and multimodal tasks
Close to GPT-5.4 performance on some benchmarks
GPT-5.4 Nano: Smaller and cheaper
Built for speed-first tasks
Best for classification, extraction, ranking
This is where things get interesting. OpenAI is pushing a multi-agent system: GPT-5.4 → handles planning + final decisions/ GPT-5.4 Mini → handles smaller tasks in parallel.
Example: Search a codebase + Review documents + Process screenshots = All done by Mini while the main model coordinates.
GPT-5.4 Mini can now understand complex UI screenshots, work across apps, and also reason over images in real-time. This makes it useful for coding assistants, and computer-use agents. And now free users get access to this setup as well.
PRESENTED BY ATTIO
Here’s how I use Attio to run my day.
Attio is the AI CRM with conversational AI built directly into your workspace. Every morning, Ask Attio handles my prep:
Surfaces insights from calls and conversations across my entire CRM
Update records and create tasks without manual entry
Answers questions about deals, accounts, and customer signals that used to take hours to find
All in seconds. No searching, no switching tabs, no manual updates.
Ready to scale faster?
AI SOURCES FROM AI FIRE
1. Complete Claude Cowork Guide: Automate Your Whole Workflow While You Sleep. Never treat Claude Cowork like a simple tool. But behind it is a 7-layer system that can run tasks, workflows, even your business, quietly in the background.
2. ChatGPT, Claude & Gemini Aren’t the Problem. Your Current AI Workflow Is. Fix it. Here’s how to externalize memory, connect AI directly to your files, and build systems that work without constant prompting.
3. 7 GPT/Claude/Gemini Prompts to Land More Interviews (Job Hunt Hack, Real Results). Apply these to fix your resume and practice for your next big meeting. I got 5 interview calls in one week by using these very simple steps.
4. One Powerful AI System Most Aren’t Building to Fix Inconsistent Outputs!? Most people blame AI for bad results. The real issue is simpler. Once you see this system, you won’t go back to normal prompting.
5. If I Had to Build a $10M AI Business Alone, This Is Exactly What I’d Do With Zero Team. What would you build if you had no team but needed $10M? Here’s the exact AI system and model. No overhead. Just tools. Real revenue.
Live Session • 10:30 PM (22:30) EDT • Mar 18, 2026 (Today)
Imagine creating Nike-level cinematic video ads and professional product sequences without a camera crew, an expensive studio, or a month of manual editing.
In this workshop, you’ll learn how to turn AI video into a real production workflow using video tools like Kling 3.0, Veo, and Sora.
If you have a product you are trying to sell or a specific brand idea, have your high-quality images or logos ready to:
🎬 Understand why most AI videos fail
🚀 Master Kling 3.0 for multi-shot videos with consistent characters
🧠 Build story-driven videos instead of random generations
⚡ Follow a professional workflow step by step
📊 When to use Kling 3.0 vs. Veo vs. Sora for your business goals
TODAY IN AI
AI HIGHLIGHTS
👀 A cybersecurity startup hacked McKinsey’s internal AI bot, 46.5M chats leaked… If this can happen to them, what about everyone else? This one’s worth reading.
🚀 Okara launched an “AI CMO” that deploys marketing agents from a single URL. It auto handles SEO, content, Reddit, X. It’s offering a promo for new users, claim here.
🚨 YC CEO Garry Tan open-sourced his Claude “gstack” AI workflow. It simulates a full dev team with agents. It went viral. Some call it genius. Others say it’s… nothing?
💰 OpenAI's “unlimited” ChatGPT plans might be ending as usage explodes. Pricing could shift toward pay-per-use, like electricity. How much will AI really cost you?
🤯 Gamma entered Canva territory with AI images. Their new “Imagine” tool turns prompts into full marketing assets. It feels like PowerPoint + AI design… but faster.
🥇 Remember Manus AI from our last superagent workshop? It just dropped “My Computer,” a desktop app that works directly on your device. Watch it in action.
⚡ GPT-5.4 exploded in just 7 days. 5 trillion tokens/day and already $1B run rate. This is way faster than any previous API record set by OpenAI. What’s really driving it?
⚠️ The Pentagon is replacing Anthropic after their deal collapsed. Now it’s building its own AI while working with OpenAI and xAI, according to a Bloomberg conversation.
💰 Big AI Fundraising: Gradient, backed by Google, has launched a $220M fund to support early-stage AI startups, with innovative technology from a major player.
NEW EMPOWERED AI TOOLS
🖥️ My Computer by Manus AI brings Manus onto your computer, letting your AI agent work directly with local files, tools, and apps.
🎬 Kira brings video and music into the same creative flow. Animate any photo, drop an AI soundtrack, change hairstyle before anyone notices.
🤖 Codex Subagents allows you to spawn specialized, parallel AI workers for complex tasks, execute multi-step workflows without context rot.
🎮 NVIDIA DLSS 5 analyzes color & motion vectors to deliver Hollywood-grade fidelity in real time, not just performance upscaling.
AI BREAKTHROUGH
To understand how a tumor fights the immune system, researchers need protein-level data. A basic tissue slide costs: $5–$10. A full protein test costs: $2,000+ per patient. Why so expensive? Because the process:
Requires rare materials
Needs complex imaging systems
Doesn’t scale well
Microsoft released GigaTIME, an AI model that can turn cheap medical images into detailed protein maps of cancer cells. It looks at a simple digital image and predicts:
Where proteins are located
How the tumor interacts with the immune system
So instead of running a physical test, you simulate it with AI. Microsoft trained GigaTIME on 40 million cells, then applied it to 14,256 patients, generated 300,000 detailed images.
→ From that, researchers found 1,234 new links between proteins and survival rates. You can now analyze huge patient populations, not just small lab samples. $2,000 test is replaced by a $5 image + AI.
We read your emails, comments, and poll replies daily
How would you rate today’s newsletter?Your feedback helps us create the best newsletter possible |
Hit reply and say Hello – we'd love to hear from you!
Like what you're reading? Forward it to friends, and they can sign up here.
Cheers,
The AI Fire Team
Reply