Page de couverture de 📆 ThursdAI - June 12 - Meta’s $15B ScaleAI Power Play, OpenAI’s o3-pro & 90% Price Drop!

📆 ThursdAI - June 12 - Meta’s $15B ScaleAI Power Play, OpenAI’s o3-pro & 90% Price Drop!

📆 ThursdAI - June 12 - Meta’s $15B ScaleAI Power Play, OpenAI’s o3-pro & 90% Price Drop!

Écouter gratuitement

Voir les détails du balado

À propos de cet audio

 Hey folks, this is Alex, finally back home! This week was full of crazy AI news, both model related but also shifts in the AI landscape and big companies, with Zuck going all in on scale & execu-hiring Alex Wang for a crazy $14B dollars. OpenAI meanwhile, maybe received a new shipment of GPUs? Otherwise, it’s hard to explain how they have dropped the o3 price by 80%, while also shipping o3-pro (in chat and API). Apple was also featured in today’s episode, but more so for the lack of AI news, completely delaying the “very personalized private Siri powered by Apple Intelligence” during WWDC25 this week. We had 2 guests on the show this week, Stefania Druga and Eric Provencher (who builds RepoPrompt). Stefania helped me cover the AI Engineer conference we all went to last week, and shared some cool Science CoPilot stuff she’s working on, while Eric is the GOTO guy for O3-pro helped us understand what this model is great for! As always, TL;DR and show notes at the bottom, video for those who prefer watching is attached below, let’s dive in! Big Companies LLMs & APIsLet’s start with big companies, because the landscape has shifted, new top reasoner models dropped and some huge companies didn’t deliver this week! Zuck goes all in on SuperIntelligence - Meta’s $14B stake in ScaleAI and Alex WangThis may be the most consequential piece of AI news today. Fresh from the dissapointing results of LLama 4, reports of top researchers leaving the Llama team, many have decided to exclude Meta from the AI race. We have a saying at ThursdAI, don’t bet against Zuck! Zuck decided to spend a lot of money (nearly 20% of their reported $65B investment in AI infrastructure) to get a 49% stake in Scale AI and bring Alex Wang it’s (now former) CEO to lead the new Superintelligence team at Meta. For folks who are not familiar with Scale, it’s a massive company in providing human annotated data services to all the big AI labs, Google, OpenAI, Microsoft, Anthropic.. all of them really. Alex Wang, is the youngest self made billionaire because of it, and now Zuck not only has access to all their expertise, but also to a very impressive AI persona, who could help revive the excitement about Meta’s AI efforts, help recruit the best researchers, and lead the way inside Meta. Wang is also an outspoken China hawk who spends as much time in congressional hearings as in Slack, so the geopolitics here are … spicy. Meta just stapled itself to the biggest annotation funnel on Earth, hired away Google’s Jack Rae (who was on the pod just last week, shipping for Google!) for brainy model alignment, and started waving seven-to-nine-figure comp packages at every researcher with “Transformer” in their citation list. Whatever disappointment you felt over Llama-4’s muted debut, Zuck clearly felt it too—and responded like a founder who still controls every voting share. OpenAI’s Game-Changer: o3 Price Slash & o3-pro launches to top the intelligence leaderboards!Meanwhile OpenAI dropping not one, but two mind-blowing updates. First, they’ve slashed the price of o3—their premium reasoning model—by a staggering 80%. We’re talking from $40/$10 per million tokens down to just $8/$2. That’s right, folks, it’s now in the same league as Claude Sonnet cost-wise, making top-tier intelligence dirt cheap. I remember when a price drop of 80% after a year got us excited; now it’s 80% in just four months with zero quality loss. They’ve confirmed it’s the full o3 model—no distillation or quantization here. How are they pulling this off? I’m guessing someone got a shipment of shiny new H200s from Jensen!And just when you thought it couldn’t get better, OpenAI rolled out o3-pro, their highest intelligence offering yet. Available for pro and team accounts, and via API (87% cheaper than o1-pro, by the way), this model—or consortium of models—is a beast. It’s topping charts on Artificial Analysis, barely edging out Gemini 2.5 as the new king. Benchmarks are insane: 93% on AIME 2024 (state-of-the-art territory), 84% on GPQA Diamond, and nearing a 3000 ELO score on competition coding. Human preference tests show 64-66% of folks prefer o3-pro for clarity and comprehensiveness across tasks like scientific analysis and personal writing.I’ve been playing with it myself, and the way o3-pro handles long context and tough problems is unreal. As my friend Eric Provencher (creator of RepoPrompt) shared on the show, it’s surgical—perfect for big refactors and bug diagnosis in coding. It’s got all the tools o3 has—web search, image analysis, memory personalization—and you can run it in background mode via API for async tasks. Sure, it’s slower due to deep reasoning (no streaming thought tokens), but the consistency and depth? Worth it. Oh, and funny story—I was prepping a talk for Hamel Hussain’s evals course, with a slide saying “don’t use large reasoning models if budget’s tight.” The day...

Ce que les auditeurs disent de 📆 ThursdAI - June 12 - Meta’s $15B ScaleAI Power Play, OpenAI’s o3-pro & 90% Price Drop!

Moyenne des évaluations de clients

Évaluations – Cliquez sur les onglets pour changer la source des évaluations.

Évaluations sur Audible.ca

Évaluations sur Amazon.ca
Il n'y a pas encore de critiques pour ce titre.