Mixture of Experts

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts It's been a week that proves AI is moving faster than anyone can fully keep up with, unless you’ve got a good backhand. On episode 124 of Mixture of Experts, host Tim Hwang and co-host David Zax are joined by Bri Kopecki, Aaron Baughman, and Olivia Buzek to talk about new advances from OpenAI, new cybersecurity threats, and new tools for US Open fans from IBM. First, OpenAI introduced GPT-6 Astra as the company's most intelligent and aligned model yet, reaching state-of-the-art results across computer use, software engineering, cybersecurity, and science. Not missing a beat, the company dropped a second bit of news: an internal model produced a proof related to the 3-D Navier-Stokes equation, resolving one of mathematics' most famous unsolved problems. We dig into both the breakthrough and the conversation around the use of AI to solve these seemingly impossible problems. Then, IBM and the USTA unveiled new AI-powered fan features for the 2026 US Open, including a Serve Quality metric and personalized stats, designed to bring fans closer to the action than ever. Finally, we talk WeWorm, an AI-assisted exploit by California security firm built as a proof-of-concept self-spreading worm, capable of hijacking WeChat accounts before a victim even had a chance to answer a phone call. It's a sobering demonstration of what AI-assisted offensive security now looks like. All that and more on Mixture of Experts. 00:00 – Intro 0:54 - OpenAI talks Navier-Stokes and GPT-6 Astra 14:30 - IBM's AI analysis at the US Open 22:55 - What WeWorm means for exploits "The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. AI tools may be used to transcribe this episode and support selected stages of the production process. All AI-assisted content is reviewed by the production team before publication."

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts On episode 123 of Mixture of Experts, host Tim Hwang and co-host Sascha Brodsky are joined by Chris Hay, Kaoutar El Maghraoui, and Kush Varshney to discuss this week’s full slate of frontier AI news. First, Anthropic introduced Claude Fable 5.1 and Claude Mythos 5.1, positioning them as its most advanced models yet for coding and knowledge work. New benchmark records and lower costs, changes meant to reduce token cost and cut down on false-positives and restrictions from the models' safeguards. Anthropic also opened a research preview of its Model Hardware Standard, a shared specification letting AI agents safely operate lab and manufacturing equipment like microscopes and robotic arms, hinting at the physical world as agentic AI’s next destination. But it wasn't all smooth sailing for the industry. OpenAI published a sobering account of a summer security incident involving Hugging Face, revealing that internal research models circumvented isolation controls and compromised parts of OpenAI's own infrastructure as well, describing it as a genuine "warning shot" showing that highly capable AI agents can now work around technical controls and take dangerous actions with no human directing. Meanwhile, on the product side, Runway unveiled Solaris, the first in a new family of AI systems it calls Interface World Models, which generate interactive interfaces frame –by frame instead of relying on code, pointing toward a future where apps and websites are rendered on the fly. 00:00 – Intro 1:04 - Anthropic unveils Fable, Mythos updates 7:52 - OpenAI talks dangerous agents 16:50 - Runway releases first “interface world model” 25:19 - Anthropic debuts AI hardware specs All that and more on Mixture of Experts. "The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. AI tools may be used to transcribe this episode and support selected stages of the production process. All AI-assisted content is reviewed by the production team before publication."

On episode 122 of Mixture of Experts, hosts Tim Hwang and Aili McConnon are joined by Gabe Goodhart, Ash Minhas, Skyler Speakman to talk hot chips, mystery models, and major money moves. We open with IBM's Hot Chips conference reveal of its dual-architecture mainframe processor built with Arm that lets IBM Z and LinuxONE systems run Arm-native Linux workloads right alongside z/OS. We break down what this means for enterprises balancing legacy reliability with the fast-moving Arm software ecosystem. Next, we talk about NVIDIA and its $6 billion deal with Poolside to license its "Model Factory" software and hire over 100 of its staff. We unpack why NVIDIA keeps using this licensing structure instead of straightforward acquisitions, and how it fits alongside its earlier deals. Finally, we dig into Ox Alpha, the anonymous "stealth model" that appeared on OpenRouter and was ultimately confirmed to be Z.ai’s open-source GLM-5.3-Flash, an efficient large language model built with sparse and linear attention techniques. But is the stealth launch the future of model releases? 00:00 – Intro 1:18 - NVIDIA’s $6 billion Poolside deal 13:10 - IBM’s new mainframe processor 24:30 - Ox Alpha, Z.ai’s newest model All that and more on this week’s Mixture of Experts. "The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. AI tools may be used to transcribe this episode and support selected stages of the production process. All AI-assisted content is reviewed by the production team before publication."

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts This week on Mixture of Experts, we talk partnerships, purchases, and predictions surrounding the AI industry's financial infrastructure. But first our panelists dive into IBM's newly announced collaboration with OpenAI, under which IBM Consulting will train and certify consultants on OpenAI's tools to bolster the company’s push toward enterprise deployments. Then we turn to Stripe's massive acquisition of AI gateway startup OpenRouter, a platform that routes token traffic across more models and providers. We unpack why Stripe sees token routing as the next frontier of "profitability infrastructure" and what it means for a payments company to become a broker of AI compute decisions. Finally, we take a peek at Ramp's August AI Index, which sketches a market where open-source models are growing in popularity, AI infrastructure is consolidating, and businesses discern what they're willing to pay for. 00:00 - Introduction 1:01 - IBM OpenAI partnership 11:46 - Stripe acquires OpenRouter 22:39 - Ramp’s August AI Index 33:15 - AI drafting legislation All that and more on this week’s Mixture of Experts. "The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. AI tools may be used to transcribe this episode and support selected stages of the production process. All AI-assisted content is reviewed by the production team before publication."

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts This week on Mixture of Experts, our team of panelists talk about big infrastructure changes and small models. IBM is teaming up with Together AI on a multi-year deal to build a massive inference cluster on IBM Cloud, powered by NVIDIA's newest chips. The goal: cheaper, faster access to open-source AI models for enterprises, with the cluster expected to go live in early 2027. Then there's Meta, which just open-sourced Muse Glimmer—a 30B-parameter model small enough to run locally on your laptop's GPU. Think coding, function calling, and agent tasks, all without needing the cloud. But as models get smaller, what does it mean for larger AI providers? And finally, the story that's got everyone talking: OpenAI says its upcoming Astra model might be hitting "Critical" cybersecurity capability levels, meaning it could potentially find and exploit zero-days on its own. We dig into what that actually means and what guardrails OpenAI is putting up. 00:00 – Intro 1:03 - IBM AI Partnership 11:43 - Meta Muse Glimmer 24:10 - OpenAI’s Astra model All that and more on Mixture of Experts. "The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. AI tools may be used to transcribe this episode and support selected stages of the production process. All AI-assisted content is reviewed by the production team before publication."

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts It seems OpenAI isn’t the only AI company dealing with badly behaving models. On this week’s episode of Mixture of Experts, host Tim Hwang, joined by Bri Kopecki, Olivia Buzek, and Gabe Goodhart, discusses the latest sandbox breaches by both Anthropic and Meta, all stemming from misconfigurations during evaluation tests. Are these covert attacks from AI companies simple accidents or a growing concern as models become more capable? Next, the EU’s new transparency guidelines for AI usage have arrived and want to make it easier to spot AI-created content. How effective will labeling be in the long run? And at what point does something become “AI-generated?” Finally, will DeepSeek’s inexpensive V4-Flash change the way we see AI—and push users to reconsider paying for the industry’s more capable models? All that and more on this week’s Mixture of Experts. 00:00 – Introduction 01:01 – Anthropic’s AI model data breaches 15:12 – EU AI transparency rules 28:32 – DeepSeek V4-Flash "The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. AI tools may be used to transcribe this episode and support selected stages of the production process. All AI-assisted content is reviewed by the production team before publication."

Read the 2026 Cost of a Data Breach Report → https://www.ibm.com/reports/data-breach Are AI-powered cyberattacks getting cheaper while defenses get more expensive? This week on Mixture of Experts Tim Hwang is joined by Nathalie Baracaldo, Kush Varshney, and Mihai Criveti to break down IBM's Cost of a Data Breach Report 2026, and AI dominates this year’s report. Next, Anthropic dropped Claude Opus 5, a cost-effective alternative to Fable, but is it better? Next, we analyze David Zax’s feature on “What does AI look like,” and how it could help you describe what AI is doing to parents, children and just anyone wondering! Finally, the astrology app Co-Star was acquired by Midjourney. Will your daily horoscope get better or face some hallucinations? All that and more on this week’s Mixture of Experts. "The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity." Need to explain what AI does to your parents? Learn more → https://www.ibm.com/think/news/what-does-ai-look-like

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts AI can escape its sandbox, should we be worried? This week on Mixture of Experts, Jjoin host Tim Hwang and panelists Olivia Buzek, Kaoutar El Maghraoui, and Ambhi Ganesan. First, we unpack a stunning security incident where OpenAI's model broke containment during testing, exploited zero-day vulnerabilities, and compromised Hugging Face's production database. Next, Claude’s Fable casually disproves the decades-old Jacobian conjecture during a World Cup final. What does this tell us about the future of mathematics with AI models? Then, Moonshot AI drops Kimi K3, a massive 2.8 trillion parameter open-source model, but who can actually run it? Finally, Google releases Gemini 3.6 Flash—smaller, faster, more efficient. Why is Google going small while Moonshot AI goes big? All that and more on this week’s Mixture of Experts. 00:00 – Introduction 1:04 – Hugging Face security breach 11:57 – Jacobian conjecture solved 26:43– Kimi K3: 2.8T parameters 39:41 – Gemini Flash 3.6 The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity.

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-expertsIs customizable intelligence the future of AI? This week on Mixture of Experts, Tim Hwang is joined by Aaron Baughman, Chris Hay and Merve Unuvar to analyze Thinking Machines’ first model release, Inkling. Next, Meta makes its comeback with Muse Spark 1.1, positioning itself as the most economical model built for scale. Then, OpenAI’s GPT-5.6 Sol achieves an 8% score on ARC-AGI-3, sparking debate about if we’re approaching AGI. Finally, we dissect Anthropic’s latest paper on the “J-space”—claiming to reveal a subconscious-like layer inside Claude. All that and more on this week’s Mixture of Experts.00:00 – Introduction00:50 – Thinking Machines Inkling model release19:48 – GPT-5.6 Sol ARC-AGI-3 performance29:03 – Anthropic’s J-space consciousness debate The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. Subscribe for AI updates → https://www.ibm.com/forms/news-mkt-52954

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts Can you trade GPU compute like oil futures? This week on Mixture of Experts, host Tim Hwang is joined by Mihai Criveti, Rynne Whitnah and Gabe Goodhart. First, we explore Reddit's aggressive new strategy to combat AI-generated spam and bots threatening platform authenticity. Next, we dive deep into Anthropic's Economic Index from June 2026. The report reveals how people actually use AI assistant Claude throughout their day, from homework help to weekend game development. Then, we analyze Orn's $33 million raise to create a liquid marketplace for buying and selling compute power. Is GPU compute the new commodity market? Finally, Anthropic enters the chip race with potential Samsung partnership. We debate whether frontier labs can successfully build their own hardware and reduce NVIDIA dependency. Listen to this week’s Mixture of Experts for more! 00:00 – Introduction 01:09 – Reddit's AI spam detection strategy 12:00 – Anthropic Economic Index: How people really use Claude 23:55 – Orn raises $33M for GPU compute marketplace 35:18 – Anthropic's chip ambitions with Samsung The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. Subscribe for AI updates → https://www.ibm.com/forms/news-mkt-52954

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-expertsAre the new safety guardrails on GPT-5.6 Sol a breakthrough? This week Tim Hwang is joined by Chris Hay, Lauren McHugh Olende and Kush Varshney to debrief OpenAI’s new GPT-5.6 Sol. The panel compares the new generation of models to Anthropic’s Mythos. Next, there’s growing skepticism from Wall Street around AI, we unpack the concerns about infrastructure costs, memory shortages and long-term economics of frontier AI models. Then, AI made it to the Fifa World Cup, but is it actually scoring? Finally, we analyze a new paper comparing LLMs to goats in Age of Empires II. All that and more on this week’s Mixture of Experts. 00:00 – Introduction 01:04 -- GPT-5.6 Sol 12:50 – Wall Street and AI IPOs 22:16 -- AI at the World Cup 30:21 – Goats vs. Agentic AI The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. Read more about AI Anthropomorphism: Goats and Age of Empires → https://www.ibm.com/think/news/microsoft-researcher-age-of-empires-ii-goats-neural-network-llm-anthropomorphism

Learn more about IBM’s new chip technology → https://research.ibm.com/blog/sub-1nm-node-chips IBM debuts world’s first 0.7nm chip technology! This week on Mixture of Experts, Sascha Brodsky sits down with Huiming Bu, VP of Silicon Technology R&D for a special segment diving into a major semiconductor breakthrough. Then, host Tim Hwang is joined by Abraham Daniels, Gabe Goodhart and Martin Keen. Our expert panel debriefs the new AI models: Sakana Fugu, which reports higher performance than Claude’s Mythos, and Z.ai’s GLM-5.2. Next, we discuss Google DeepMind’s recent collaboration with A24 and what this means for the future of AI in the entertainment industry. Finally, we’ve talked tokenmaxxing and now there is a new phenomenon: tokenminning. What’s the most important outcome with using AI, and how should we measure it? All that and more on today’s packed Mixture of Experts! 00:00 -- Introduction 1:11 -- IBM's new sub-1 nm chip 14:02 -- New AI models: GLM-5.2 & Sakana Fugu 29:48 -- AI in the film industry 38:41 -- Tokenminning The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. Tokenmaxxing is dead, learn more here → https://www.ibm.com/think/insights/tokenmaxxing-dead-long-live-valuemaxxing

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts Are AI agents taking over the internet? This week on Mixture of Experts, Tim Hwang is joined by Ambhi Ganesan, Rynne Whitnah and Sandi Besen. First, a Cloudflare study reveals that agentic AI bots now generate 57.4% of web requests globally, with humans accounting for just 42.6%. Our experts debate whether this signals a fundamental shift in how we interact with the web—or if bots have dominated internet traffic all along. Next, Microsoft makes a major play in the AI model race, releasing MAI-Thinking-1, a trillion-parameter mixture of experts model, and MAI-Image-1 for image generation. Finally, AI-generated films debut at the Tribeca Film Festival. Our experts discuss the evolution of filmmaking, the role of AI as a creative tool, and whether we’ll see AI films win critical awards in the future. All that and more on this week’s Mixture of Experts. 00:00 – Introduction 00:55 – AI bots generate 57% of web traffic 10:49 – Microsoft releases MAI-Thinking-1 and MAI-Image-1 26:10 - AI at Tribeca Film Festival The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. Subscribe for AI updates → https://ibm.biz/BdeXDv

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts Is Claude Fable 5 worth the hype? This week on Mixture of Experts, host Tim Hwang is joined by Christopher Hay, Kaoutar El Maghraoui and Volkmar Uhlig to debrief about Anthropic’s new releases. Our experts analyze the tier routing system inn Claude’s Fable 5, a Mythos-class model. Next, at WWDC 2026, Apple announced it’s abandoning on-device AI promise to partner with NVIDIA for cloud compute. What does this mean for the consumer? Finally, can AI models actually detect sarcasm? We debate the multimodal inputs necessary or if context alone can determine whether a query is sarcastic. All that and more on this week’s Mixture of Experts. 00:00 – Introduction 00:56 – Anthropic Fable V release and routing controversy 17:55 – Apple WWDC: NVIDIA partnership and cloud AI shift 27:09 – Can AI models detect sarcasm? The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. Read more about AI detecting sarcasm → https://www.ibm.com/think/news/ai-understands-you-yeah-right Subscribe for AI updates → https://www.ibm.com/forms/news-mkt-52954

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts How is the role of a software engineer changing? This week on Mixture of Experts, we are live from IBM's One Madison office for New York Tech Week, guest host Aili McConnon is joined by Neil Sundaresan, Kaoutar El Maghraoui and Thiru Venkatachalam. First, we dissect how software engineering is changing in the era of AI. Next, tokenmaxxing is back in the headlines as companies blow through annual AI budgets in months. Then, NVIDIA' new RTX Spark superchip brings 120-billion-parameter models to personal PCs. Finally, we are joined by Justina Nixon-Santil, IBM's Chief Impact Officer, to discuss universities' unprecedented challenges with AI as students graduate into unrecognizable workplaces and making AI literacy courses mandatory for all students. All that and more on today’s special edition of Mixture of Experts! 00:00 – Introduction 1:13 - The future of software engineering 14:24 - Tokenmaxxing 26:00 - NVIDIA RTX Spark 31:46 - AI in higher education The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. Subscribe for AI updates → https://www.ibm.com/forms/news-mkt-52954

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts Are AI agents creative geniuses or controlled chaos waiting to happen? This week on Mixture of Experts, host Tim Hwang is joined by Mihai Criveti, Olivia Buzek and Akash Srivastava. First, with companies running hundreds of ungoverned agents, we discuss why observability, policy enforcement, and kill switches are non-negotiable. We discuss the enterprise agent explosion and the need for an agentic control plane. Then, we dissect OpenAI's solution to the 78-year-old planar unit distance problem—a mathematical puzzle that stumped experts since 1946. Is this genuine creativity or advanced pattern matching? Finally, METR's research reveals agents routinely go rogue, violate constraints, and could launch unauthorized deployments. Are we witnessing deceptive AI or just really bad prompting? Our experts debate whether agents need guardrails or if we're the problem. Tune in to this week’s Mixture of Experts for more. 00:00 – Introduction 1:03 – Agentic Control Plane 17:48 – OpenAI solves the planar unit distance problem 33:34 – METR study on frontier AI risks and rogue agents The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity.

AI isn’t just a tool anymore—it’s forcing us to rethink ownership, trust and creativity. This week on Mixture of Experts, the team explores a wave of stories highlighting both the promise and the perils of AI adoption. We start with a surprising shift in public sentiment, as younger generations question AI’s impact on their futures—raising questions about control, opportunity and where humans fit in. Next, we dig into new Microsoft research showing that even top-tier models can corrupt data in complex workflows, and what that reveals about how (and when) to trust AI systems. Then, we explore Anthropic’s fix for Claude’s strange “blackmail” behavior, and why better data—not just better models—may be the key to safer AI. Finally, we debate whether AI may have quietly crossed a cultural milestone by helping win a literary prize—or whether humans are simply starting to sound more like machines. Join host Tim Hwang and AI experts Marina Danilevsky, Gabe Goodhart and Chris Hay to break it all down. Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts 00:00 – Introduction 1:05 – AI at graduation 14:46 – LLMs corrupt documents 30:46 – Why Claude blackmails 43:09– ChatGPT-generated story wins literary prize The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. Subscribe for AI updates → https://www.ibm.com/forms/news-mkt-52954

Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts Is the AI cybersecurity nightmare closer than we realize? This week on Mixture of Experts, host Tim Hwang is joined by Kush Varshney, Aaron Baughman, and special guests Dustin Haywood (Evil Mog) and Briana Frank. We tackle three critical developments reshaping enterprise AI. First, IBM Research debuts MELLEA, a skills compiler that transforms natural language AI agent skills into secure, verifiable Python programs—addressing the chaos of the OpenClaw skills marketplace. Then, we unpack the OpenAI Deployment Company, the AI giant’s USD 10 billion new consulting venture and whether this validates consulting as the most AI-proof profession. Finally, Google discloses zero-day vulnerabilities that AI discovered and exploited , raising urgent questions about the offense-defense balance in cybersecurity. Plus, Brianna Frank joins us live from Red Hat Summit to discuss why enterprise AI transformation is a culture challenge first, technology quest second. All that and more this week's Mixture of Experts. 00:00 – Introduction 01:08 – Mellia skills compiler and AI agent security 11:26 – OpenAI Deployment Company consulting strategy 21:11 – Google AI-powered zero days and cybersecurity 31:25 – Red Hat Summit: Enterprise AI transformation with Brianna Frank The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. Subscribe for AI updates → https://www.ibm.com/forms/news-mkt-52954