- The Rundown AI
- Posts
- OpenAI's 'cheap' new frontier model
OpenAI's 'cheap' new frontier model
PLUS: Meta launching ‘superintelligence’ lab with Scale AI founder
Good morning, AI enthusiasts. OpenAI just dropped o3-pro, the company’s new top reasoning model — coming alongside a massive price cut that shows intelligence is becoming cheaper than ever, even as capabilities soar.
With early testers calling the release "much smarter" and CEO Sam Altman writing that humanity has “passed the AI event horizon,” the optimism coming out of the AI leader feels higher than ever.
In today’s AI rundown:
OpenAI’s o3-pro with massive price cuts
Meta’s ‘superintelligence’ lab with Scale AI founder
How to create AI videos for free using OpenAI's Sora
OpenAI CEO says AI takeoff has started
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
OPENAI

Image source: OpenAI
The Rundown: OpenAI released o3-pro, an upgraded version of its reasoning model that outperforms competitors on key benchmarks — while simultaneously reducing its o3 prices by 80% in a direct challenge to Google and Anthropic’s top models.
The details:
o3-Pro is designed to "think longer," boosting reliability and performance in technical fields like math, science, and programming.
The model outperforms top rivals on PhD-level math and science tasks, with evaluators preferring o3-pro across all tested categories.
It can also use tools like web search and data analysis, but is slower and lacks support for features like image generation and Canvas.
ChatGPT Pro and Team users gain immediate access, with Enterprise and Edu customers receiving the model next week.
Why it matters: According to tester Ben Hylak, o3-pro feels “much smarter” than its predecessor and “very different” from other frontier models. But the most mind-blowing part of this release may be the pricing — coming in at a fraction of older models despite a gigantic leap in intelligence and capability.
TOGETHER WITH AUGMENT CODE
The Rundown Augment Code's powerful AI coding agent meets professional software developers exactly where they are, delivering deep context and production-ready capabilities. And now Augment Code Remote Agents are available in VS Code to help your team clear out its backlog.
While you and your team build what’s next, Remote Agent can:
Tackle technical debt
Refactor code
Update tests and docs
Sign up today and let Augment Code transform how your team works.
META

Image source: Getty Images / Reuters
The Rundown: Meta is reportedly shaking up its AI division with a new “superintelligence lab” hand picked by CEO Mark Zuckerberg, with a multi-billion dollar deal to bring Scale AI CEO Alexandr Wang and other top talent onboard.
The details:
Wang will lead Meta's new group alongside other Scale AI technical talent, with the 28-year-old founder taking a top position in Zuckerberg's AI hierarchy.
The $15B deal sends cash to Scale's existing shareholders while allowing Meta to sidestep regulatory acquisition concerns with a 49% stake in the company.
Zuckerberg has personally recruited nearly 50 researchers for the lab, offering as high as nine-figure packages to poach talent from OpenAI and Google.
The move follows Zuckerberg’s reported frustration with the performance of Meta’s Llama 4 model and a desire to accelerate past competitors.
Why it matters: After drama and disappointment surrounding the recent Llama 4 release, Zuck is taking matters into his own hands — and isn’t afraid of shaking up his existing staff and breaking the bank to do it. Meta has the resources, but a dramatic pivot might be what’s truly necessary to catch up to its powerhouse AI rivals.
AI TRAINING

The Rundown: In this tutorial, you will learn how to use OpenAI’s powerful Sora video generator through Microsoft’s Bing mobile app to create professional-quality videos — instantly and without any cost.
Step-by-step:
Download the Microsoft Bing app and select the mini-apps icon in the bottom right corner.
Access the “Video Creator” feature from the available mini-apps
Write a detailed prompt describing your desired video scene
Generate your 5-second, 9:16 format video and download via the three dots menu
Pro tip: Create multiple variations of the same concept by slightly adjusting your prompts.
PRESENTED BY ANGELQ
The Rundown Meet AngelQ: the first AI-powered super browser made just for kids. Designed to turn summer screen time into a launchpad for learning—with built-in safety every step of the way.
AngelQ’s safe, age-appropriate platform:
Teaches kids how to use AI the right way — responsibly and confidently
Acts as a protected space for exploring, searching, and discovering
Helps children build real tech skills they’ll carry into the future
Is designed to grow with kids as they progress
Give your child a smart start to AI with a free year of AngelQ by using code RUNDOWN365.
SAM ALTMAN

Image source: o3 / The Rundown
The Rundown: OpenAI CEO Sam Altman published a new blog post titled “The Gentle Singularity,” claiming humanity has passed the AI event horizon — predicting superintelligence will reshape society, but its changes will remain manageable.
The details:
Altman frames the takeoff as a “gentle singularity,” where society adapts to exponential progress as once-amazing capabilities quickly become routine.
His timeline includes the AI creating new ideas in 2026, robots functioning in the real world in 2027, and an explosion of creation across industries.
In the 2030s, he projects that both intelligence and energy will become abundant, with the cost of AI eventually approaching the cost of electricity.
Altman’s path forward involves first solving AI alignment, then ensuring superintelligence is cheap, distributed, and not controlled by a single entity.
Why it matters: Despite AI frequently being painted in doomsday scenarios, the man at the center of it all envisions a quite different outlook. Users already quickly adapt to each new wave of intelligence, moving on from one release to shouting for the next — and Altman sees that continuing even as we cross a new threshold of tech advances.
QUICK HITS
📚 Exa Research Pro - New SOTA agentic tool for web research
🍎 Foundation Models - Build on Apple’s on-device Apple Intelligence model
⚙️ Xcode 26 - Apple’s coding tool, with new access to AI models
🗂️ Common Pile v0.1 - 8TB dataset of public and licensed text for AI training
🔁 The Rundown - Strategic Partnerships (AI University)
🧑🏫 Parloa - Senior Product Trainer
🌟 Synthesia - Director, Customer Success
🌍 Palantir Technologies - Global Mobility Specialist
Mistral released Magistral, its open-source reasoning family with quick responses and multi-language support, though STEM and coding benchmarks lag behind top rivals.
OpenAI finalized a deal with Google Cloud for additional compute, diversifying beyond Microsoft as Google expands its cloud business to include its biggest AI competitor.
Google added a new Veo 3 Fast version of its viral video generation model in Gemini and Flow, allowing for expanded access with 2x the speed.
KREA AI unveiled Krea 1, the company’s first in-house image model — launching in free beta with enhanced aesthetic control, artistic knowledge, and image quality.
Enterprise AI startup Glean raised $150M in new funding at a $7.2B valuation, driven by adoption from Fortune 500 companies and its Glean Agents platform.
SAG-AFTRA and major video game companies reportedly reached a tentative deal to end a nearly year-long strike by actors over AI and compensation protections.
COMMUNITY
Check out our last workshop with Dr. Alvaro Cintas, The Rundown’s AI professor. By the end of the workshop, you’ll confidently be able to incorporate Flux Kontext into your creative workflows for consistent, high-quality results.
Watch here. Not a member? Join The Rundown University on a 14-day free trial.
That's it for today!Before you go we’d love to know what you thought of today's newsletter to help us improve The Rundown experience for you. |
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
Reply