• AI Fire
  • Posts
  • ⚡ Everyone Shipped At Once

⚡ Everyone Shipped At Once

Google. xAI. OpenAI. Perplexity

In partnership with

ai-fire-banner

Google shipped. xAI shipped. OpenAI shipped. Then researchers dropped results that rewrote what models were supposed to handle.

IN PARTNERSHIP WITH BELAY

Q4 is the perfect window to turn this year’s numbers into a clear, actionable forecast aligned with your goals. Set your business up for a stronger 2026 with BELAY’s new guide.

AI INSIGHTS

gemini-3-pro-beats-rad-trainees

For the first time, a general AI model outscored radiology trainees on RadLE v1, a tough 50-case exam built to mimic real hospital complexity.

Here’s the jump:

  • Radiologists: 83%

  • Gemini 3.0 Pro: 51 - 57%

  • Trainees: 45%

  • GPT-5 (reasoning-high): 30%

This matters because until now, every frontier model failed this benchmark. GPT-5, Gemini 2.5, Claude Opus 4.1, o3… all scored below trainees.

What changed: Gemini 3.0 Pro shows early signs of structured, radiologist-like reasoning.

In an appendicitis case where GPT-5 fell apart - misreading anatomy, bouncing between wrong diagnoses - Gemini did the opposite:

  • Found the appendix

  • Listed key imaging signs

  • Ruled out mimics

  • Gave one clear answer: Acute appendicitis

Three API runs showed the same pattern (57% avg.), so this wasn’t luck.

Why it matters: Generalist models are finally breaking into real medical reasoning. But they’re still far from experts. No autonomy. No clinical use. Not yet. Still, this is the biggest jump on RadLE so far. RadLE v2 comes next with more cases and sharper scoring. The curve is rising faster than many thought.

PRESENTED BY ROKU

Find your customers on Roku this Black Friday

As with any digital ad campaign, the important thing is to reach streaming audiences who will convert. To that end, Roku’s self-service Ads Manager stands ready with powerful segmentation and targeting options. After all, you know your customers, and we know our streaming audience.

Worried it’s too late to spin up new Black Friday creative? With Roku Ads Manager, you can easily import and augment existing creative assets from your social channels. We also have AI-assisted upscaling, so every ad is primed for CTV.

Once you’ve done this, then you can easily set up A/B tests to flight different creative variants and Black Friday offers. If you’re a Shopify brand, you can even run shoppable ads directly on-screen so viewers can purchase with just a click of their Roku remote.

Bonus: we’re gifting you $5K in ad credits when you spend your first $5K on Roku Ads Manager. Just sign up and use code GET5K. Terms apply.

AI SOURCES FROM AI FIRE

1. Google's Gemini 3 Pro: 1M tokens & thinking AI is now FREE. See how it handles demanding writing, visual puzzles, custom apps,…

2. No website, no team, here’s my exact path to a $1,200 AI client now. Learn the simple manual method with zero experience or fame

3. The "Free Work" AI strategy that made me $90k/month. I’ll show you the 3-step AI framework I used to scale by selling outcomes, not just time

TODAY IN AI

AI HIGHLIGHTS

🍏 Perplexity just shipped full document-creation inside search. You can make slides, sheets, and docs right in the web app.

⚡ xAI dropped Grok 4.1 Fast, their new tool-calling model with a 2M context window. Paired with the Agent Tools API, it handles real workflows like booking upgrades, code execution, and multihop research.

🏛️ The White House quietly paused a draft executive-order that aimed to override state AI laws. Both parties pushed back hard. This pause affects everything from fraud rules to broadband funding tied to AI compliance.

💬 Reddit is buzzing again. Many say Gemini 3 feels like the real GPT-5. Users claim it dominates multimodal tasks and even tops humanities leaderboards without tools.

🖼️ OpenAI now shows more inline-images in ChatGPT answers. People, places, products. Cleaner context in one glance. Rolling out on web, iOS, and Android for GPT-5.1 responses.

💰 AI Daily Fundraising: Genspark raised $275M at a $1.25B valuation to scale its new AI Workspace. The round includes Emergence Capital, SBI Investment, LG Electronics, Pavilion Capital, and Uphonest Capital. The suite runs on 30+ AI models, including Google’s new Nano Banana Pro.

Search Party reveals exactly what AI tells buyers about your brand.
Then helps you control those crucial answers.

  1. Google Antigravity’s tutorial on building a self-playing pinball machine using Antigravity as an agentic development platform

  2. Replit’s step-by-step guide to building an AI Company Researcher web app with Replit, Serper, and OpenAI

  3. Google Cloud’s walkthrough on trying Gemini 3 Pro in Vertex AI Model Garden with 200+ foundation models and one-click deployment

  4. Replit’s hands-on guide to building an AI YouTube Comment Manager bot that replies to comments based on your video content

NEW EMPOWERED AI TOOLS

  1. 🕵️ Browser Cash is the stealth browser automation platform for scraping, browsing & data extraction with zero CAPTCHAs.

  2. ⚙️ Lamatic 3.0 helps you turn domain expertise into agentic apps, now with AgentKit, Studio & GitHub control.

  3. 📱 Comet for Android is the first agentic AI browser for mobile - summarize, chat, and browse hands-free.

  4. 📄 OCR Arena lets you compare top OCR & VLM models, test accuracy, and vote on a live public leaderboard.

SCIENCE x AI

gpt5-early-science-acceleration

Early tests show GPT-5 helping researchers across biology, math, and optimization - not just summarizing, but pushing real progress.

Biology: Months of confusion on an immune-cell shift. GPT-5 spotted the mechanism in minutes and suggested the experiment that confirmed it.

Math: On a decades-old Erdős problem, GPT-5 delivered the missing idea that completed the proof.

Optimization: For robotics and routing, GPT-5 found a failure case no one had shown and offered a cleaner proof for a recent theorem.

Overall: scientists set the direction, GPT-5 explores fast, catches mistakes, links ideas across fields, and drafts proofs quickly.

It still hallucinates and needs oversight. But it’s already shrinking weeks of work into minutes.

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

Login or Subscribe to participate in polls.

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

or to participate.