--- title: "IBM and Together AI on IBM Cloud multi-year inference cluster; go live early 2027" sdDatePublished: "2026-08-15T05:10:00Z" source: "https://www.ibm.com/think/podcasts/mixture-of-experts/ibm-cloud-meta-muse-glimmer-openai-astra-model" topics: - name: "artificial intelligence" identifier: "medtop:20001298" - name: "computer security" identifier: "medtop:20000229" - name: "computer and telecommunications hardware" identifier: "medtop:20000226" - name: "business strategy and marketing" identifier: "medtop:20000192" - name: "business restructuring" identifier: "medtop:20001366" locations: - "India" --- IBM and Together AI on IBM Cloud multi-year inference cluster; go live early 2027 IBM’s cloud collab, Meta’s Muse Glimmer & OpenAI’s upcoming Astra model | Mixture of Experts | IBM IBM’s cloud collab, Meta’s Muse Glimmer & OpenAI’s upcoming Astra model 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. All that and more on Mixture of Experts. 1:03 - IBM AI Partnership 11:43 - Meta Muse Glimmer 24:10 - OpenAI’s Astra model 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. Learn more about artificial intelligence Host and speakers of the episode: Anthropic’s AI models breach their sandbox, the EU announces new AI transparency rules, and DeepSeek’s V4-Flash undercuts its competition on price. Tune in to Mixture of Experts. IBM's Cost of a Data Breach report reveals AI's growing role in cybersecurity. Plus, Claude Opus 5, what AI really looks like, and Midjourney's astrology move. OpenAI’s AI breaks containment hacking Hugging Face, Fable solves the Jacobian conjecture, Kimi K3’s 2.8T parameters, and Gemini Flash 3.6. Tune in now! Thinking Machines’ Inkling, Meta’s Muse Spark 1.1, GPT-5.6 Soul’s 8% ARC-AGI-3 score, and Anthropic’s J-space. Tune into this week’s Mixture of Experts. Reddit fights AI spam, Anthropic Economic Index reveals AI usage patterns, and Orn raises USD 33M for GPU compute marketplace. Plus, Anthropic’s chip ambitions. OpenAI's GPT-5.6 Sol, Wall Street's growing AI concerns, FIFA's AI ambitions, and what goats can teach us about LLMs. IBM debuts world’s first 0.7nm chip technology! AI bots now generate 57% of web traffic. Microsoft releases MAI-Thinking-1 and MAI-Image-1 models. Plus, AI films debut at Tribeca. Tune in to MOE! Anthropic's Fable 5 release sparks debate on AI routing, Apple shifts to cloud AI with NVIDIA and can AI models detect sarcasm? Tune in to Mixture of Experts. How is AI transforming software engineering and higher education, tokenmaxxing debates, NVIDIA RTX Spark's edge computing breakthrough and preparing students for AI jobs. IBM's wastonx Agentic Control Plane tackles enterprise AI chaos, OpenAI cracks a 78-year math problem, and METR reveals frontier AI risks. Tune in now! AI at college graduations, LLMs corrupt documents, why Claude blackmails and ChatGPT-generated story wins literary prize. Tune in to this week’s Mixture of Experts podcast. MELLEA skills compiler secures AI agents, OpenAI launches consulting business and Google discovers AI-powered zero days. The cybersecurity arms race is here. Live at IBM Think 2026, Mixture of Experts dives into IBM’s AI operating model, the latest CEO study, and the economics of scaling AI. IBM launches Granite 4.1 & Project Bob for enterprise AI. Plus, DeepMind's DLOCO training, DeepSeek v4, and quantum computing's path to advantage. Apple's new CEO, customer intent in search, and Anthropic's chip partnership. Tune in to this week's Mixture of Experts podcast. Claude Opus 4.7, Apple's AI glasses strategy, workplace AI adoption stats and DeepMind's manipulation research. Tune in to this week's Mixture of Experts. Is your infrastructure safe from next-gen AI? Anthropic's Glasswing and Mythos model, OpenAI and Anthropic financials, GPT-1900 tests, Masters Vault demo. SpaceX files for IPO with AI infrastructure vision, Bluesky's AI bot gets blocked and cognitive offloading vs. surrender. Tune in to Mixture of Experts. AI agents transform real estate, scientific research and enterprise finance. Episode 100 explores ChatGPT home sales, Claude Code adoption and Adobe's AI lab. NVIDIA GTC showcases NemoClaw, Anthropic launches new institute, Shopify debuts shopping agents & OpenAI pivots to enterprise and code. OpenAI launches Codex Security, AI agents mine crypto autonomously, and eval awareness threatens benchmarks. Tune into this week’s Mixture of Experts. Perplexity Computer launches agent orchestration, Claude imports memory, NullClaw minimizes AI agents & Tilly Norwood becomes the first AI actor. AI as builder vs tool, mainframe modernization with COBOL, global AI adoption barriers, and securing AI agents through agent ops. Google's USD 200B AI infrastructure deal in India, Claude builds a C compiler in 2 weeks, AI agent security vulnerabilities and the enterprise ROI problem. Microsoft's Copilot usage report reveals surprising AI adoption trends, plus Ralph Wiggum prompting, India AI Summit and AI's Super Bowl advertising blitz. Anthropic's Claude Opus 4.6 and OpenAI's GPT-5.3-Codex dropped within an hour. We break down the enterprise AI battle and what it means for developers. OpenAI Codex app launches for coding agents while Moltbook – the agenty-only social- network exposes security risks. Tune in to this week’s Mixture of Experts! Moltbot (formerly Clawdbot) sparks open-source agent revolution, Dario Amodei's AI adolescence essay, IBM Grammy IQ and Microsoft's Maia 200 chip unveiled. OpenAI introduces ads in ChatGPT, Claude Code has its breakout moment and IBV’s Enterprise 2030 report reveals the future of AI innovation. Listen to today’s Mixture of Experts! Claude Cowork brings AI agents to your desktop, Apple switches to Google Gemini for Siri, and Linus Torvalds embraces vibe coding. All that and more on this week’s Mixture of Experts! NVIDIA's Rubin promises 5X performance gains, Meta acquires Manus AI for USD 2 Billion, and DeepSeek's MHC architecture redefines model training efficiency. Our experts review 2025's AI breakthroughs and predict 2026 trends. AI hardware scarcity, open source wins, super agents and multimodal evolution discussed. Take the next step in your AI Journey Easily design scalable AI assistants and agents, automate repetitive tasks and simplify complex processes with IBM watsonx Orchestrate. Put AI to work in your business with IBM’s industry-leading AI expertise and portfolio of solutions at your side. In this comprehensive guide, you will find a collection of AI agent-related content such as educational explainers, hands-on tutorials, podcast episodes and much more.