OpenAI Drops o3 Pricing 80% and Launches o3-pro

Also, Zuckerberg recruiting new team to build Superintelligence at Meta

Author

Jack Lajoie
June 11, 2025 • read time ~ 9 minutes

⚡️ Headlines

🤖 AI

GenAI for Government Transformation – AWS outlines how generative AI can streamline government operations, enhance service delivery, and reduce inefficiencies. [WSJ]

Meta Opens New AI Lab to Pursue Superintelligence – Meta has launched an advanced AI research lab aimed at developing artificial superintelligence, signaling a new strategic focus. [NYTimes]

Google Offers Deals to News Publishers to Use Its AI Tools – Google is proposing financial incentives to publishers in exchange for using its AI tools for content creation. [WSJ]

OpenAI Delays Release of Its First Open-Source Model – OpenAI has postponed the launch of its much-anticipated open-source AI model, citing alignment and safety concerns. [TechCrunch]

Mistral Unveils ‘Magistral’, a Customizable Open-Weight AI Model – Mistral introduces Magistral, a new AI model designed for enterprise use, emphasizing customization and transparency. [Mistral]

Amazon Launches AI-Powered Video Generator for Advertisers – Amazon debuts a generative AI video tool to help brands quickly create advertising content. [The Verge]

OpenAI Signs Major Cloud Deal with Google Despite Rivalry – OpenAI has inked a significant cloud computing deal with Google, defying competitive tensions. [Reuters]

Amazon Says Alexa Plus Will Reach Over a Million Users This Summer – Amazon projects its subscription-based Alexa Plus will exceed one million users by summer 2025. [The Verge]

Chinese Tech Firms Suspend AI Tools to Prevent Exam Cheating – Leading Chinese AI companies are freezing tools like chatbots to stop students from cheating during national exams. [The Guardian]

Worldcoin’s Iris-Scanning Tech Expands to the UK – Worldcoin is expanding its biometric identity system, based on iris scans, to new locations in the UK. [Bloomberg]

🦾 Emerging Tech

Snap Plans to Release Lightweight AR Glasses for Consumers in 2026 – Snap aims to launch a more affordable, user-friendly AR glasses product by 2026. [TechCrunch]

Global Robotaxi Market Gains Traction with Waymo and Apollo Go – Companies like Waymo and Apollo Go are expanding autonomous taxi services across major cities worldwide. [Rest of World]

Nvidia Sees $50 Trillion Opportunity in Physical AI Systems – Nvidia forecasts a massive market potential for AI-integrated robotics and real-world systems. [VentureBeat]

🤳 Social Media

Threads Begins Testing Direct Messaging Feature – Meta’s Threads is rolling out a direct messaging feature to select users as it builds out core functionality. [The Verge]

X’s Hardline Ad Strategy: “Advertise or We’ll Sue” – X is pressuring major advertisers through lawsuits to resume or boost ad spending amidst antitrust claims against alleged coordinating boycotts. [Wall Street Journal]

Social Media Creators to Overtake Traditional Media in Ad Revenue – Creator-driven content is projected to surpass traditional media in global ad revenue by the end of 2025. [The Guardian]

YouTube Added $55B to US GDP and 490K Jobs in 2024 – YouTube reports its ecosystem generated nearly half a million jobs and $55 billion in U.S. economic impact last year. [TechCrunch]

🔬 Research

Your Brain on ChatGPT: Accumulation of Cognitive Debt – This 206-page study analyzes how using ChatGPT for essay writing may lead to "cognitive debt," where the ease the AI provides ultimately hampers users’ critical thinking and writing skills. [arXiv]

🎱 Random

SAG-AFTRA Reaches Tentative Deal with Video Game Companies – Voice actors may soon return to work as SAG-AFTRA strikes a tentative agreement with major video game publishers. [Variety]

Samsung Adds AI Voice Recognition to Bespoke Smart Fridges – Samsung's updated smart fridges will include AI-enhanced voice features via Bixby, targeting connected home users. [The Verge]

Roblox Appoints Paramount’s Naveen Chopra as New CFO – Roblox has hired Naveen Chopra from Paramount to lead its financial strategy as Chief Financial Officer. [Bloomberg]

🔌 Plug-Into-This

OpenAI slashed prices of its o3 model by 80% and introduced o3‑pro, offering significantly improved reasoning when provided enough context. Early adopters like Ben Hylak report a 64% human preference win rate over o3, but it requires more context to shine.

  • Pricing now aligns with GPT‑4.1 ($20/$80), with o3‑pro supporting 10× base model calls and majority‑voting logic.

  • In benchmarks, o3‑pro modestly outperforms o3 on reliability metrics (4/4 tests).

  • Users note it's "pushing into uncharted territory," but benchmarking challenges make quantifying gains difficult.

  • Its performance improvements are most visible when supplied with plenty of background information.

  • Some community voices caution that high expectations make real-world evaluation ambiguous, and tasks that stump humans also stump models.

🤔 While o3‑pro marks a clear step up, its full potential depends on prompt crafting—and benchmarking remains an art more than a science. o3‑pro can think more deeply but you need to ask it smartly and load it with context for best results.

Mark Zuckerberg is personally leading the formation of a ~50‑member “Fantastic 50” superintelligence team at Meta, part of a ~$14–15 billion investment in Scale AI and to directly steer AGI development.

  • The investment secures Meta a 49% non‑voting stake in Scale AI, valued around $14.8B, with Scale’s CEO Alexandr Wang slated to join Meta.

  • Zuckerberg is offering compensation up to nine‑figure sums to poach top AI researchers.

  • The move follows Meta's frustration with delayed launches and mixed reviews of its Llama models and “Behemoth.”

  • The mini‑acquisition parallels rivals’ strategies: Amazon with Anthropic, Microsoft with OpenAI, Google with Anthropic.

  • This setup skirts antitrust issues by using non‑voting stakes while securing strategic influence.

💰 Zuckerberg is betting big—literally—to snap up talent and take control of AGI development for Meta. This aggressive hybrid of investment and headhunting shows Meta pivoting from internal R&D to external talent acquisition, a big gamble to reclaim footing in the rapidly progressing AGI race.

Sam Altman envisions a future where technological singularity unfolds gradually and humanely, emphasizing alignment, equity, and manageable progress even as AI systems approach superintelligence.

  • Altman frames the singularity as a "slow merge," a smooth continuation of exponential change that feels stable in hindsight.

  • He illustrates future milestones: breakthroughs in physics, material science, space, and brain–computer interfaces between now and 2035.

  • Alignment is central: AI must reliably reflect collective long-term goals rather than optimize short-term triggers like social media algorithms.

  • Altman stresses broad access to superintelligence to avoid economic concentration and promote shared prosperity.

  • He acknowledges risks—technical and societal—but believes careful planning can produce a “gentle singularity.”

🔭 Think of AI's future as a tricycle evolving into a flying car, not a sudden warp. Even as it transforms everything, according to Altman, the process needs to unfold step by step. Altman’s vision marries ambition with caution—but delivering on alignment and equitable distribution will demand more than good intentions.

 🆕 Updates

📽️ Daily Demo

🗣️ Discourse