– Running on Mac Studio M2 Ultra (3 years old) – Gemma 4 26B A4B Q8_0 (full quality) – Built-in WebUI (ships with llama.cpp) – MCP support out of the box (web-search, HF, github, etc.) – Prompt speculative decoding
Let me demonstrate the true power of llama.cpp: – Running on Mac Studio M2 Ultra (3 years old) – Gemma 4 26B A4B Q8_0 (full quality) – Built-in WebUI (ships with llama.cpp) – MCP support out of the box (web-search, HF, github, etc.) – Prompt speculative decoding The result: 300t/s (realtime video)
NEW EPISODE: @jack & @roelofbotha unpack @blocks 40% staff cut and rebuilding the entire company as a mini-AGI. This isn’t “use AI to make people more productive.” It’s making the company itself the intelligence. If you’re a founder or operator wondering what work looks like in the next 5 years… this is the episode. The evolution looks like: • Manager mode = Pyramid 🔺 (command & control) • Founder mode = Flat ➖(founders decide fast) • Dorsey mode = Circle 🔵 w/ AI at the center, humans at the edge, and decisions flow from customer inputs → AI → humans steering it I’ve tried killing org charts before. Brutally hard. But we never had these tools. This is rewriting the CEO playbook for the AI era. Buckle up. 00:00 Existential Dread & Hope 02:56 AI Replaces Hierarchy 07:22 Block’s New Three Roles 26:47 Flattening the Company, Fast 35:23 Getting the Board to Buy-In, Fast 36:50 Building a Great Board 41:29 Founder CEO Lessons 48:18 Second Acts & Conviction 56:22 Timeless CEO Traits
Microsoft now has MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2 available on MAI Playground!
– "MAI-Voice-1 is our top-tier voice generation model. Built to generate natural, realistic speech, rich with nuance, emotional range, and expression that preserves speaker identity… https://t.co/FQu5QncWNfpic.twitter.com/GVM0kWW76G
— 🚨 AI News | TestingCatalog (@testingcatalog) 2 avril 2026
Microsoft now has MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2 available on MAI Playground! – "MAI-Voice-1 is our top-tier voice generation model. Built to generate natural, realistic speech, rich with nuance, emotional range, and expression that preserves speaker identity
Most coding agents do not fail because they are weak. They fail because they are hard to inspect. The real problem with coding agents is not autonomy. It’s easy to make them “autonomous”. The problem is observability. A lot of tools still look impressive right until the moment they say “done”, move on, and when you check, the thing is half-built, wrong, or never happened. It happens to me almost every day, and if you don’t check for it, you might as well skip half your to-do tasks… That is why I care so much about observability and control in agentic coding. Not just more tool calls. Not just more agents. Not just more autonomy. I want to see the diff. I want to review the exact line and ensure it was done, and understand how. I want to send (only relevant) feedback back into the context. I want to compare models on a real task in my repo instead of guessing. That is what I found interesting in the rebuilt Kilo Code extension on VS Code. Yes, the parallel subagents and tons of features are cool. But the part I care about more is the (human) review loop around them. You can inspect what each agent changed, comment directly on the diff, and send those comments back as structured context. That matters. Because the value of these tools is not just in generation. It is correction. It is debugging weird hallucinations (and other LLM weaknesses). It is catching the moments where the model says “I made it” and absolutely did not. And honestly, model comparison on real tasks is underrated too. Benchmarks are nice. Your repo and actual use case are way nicer. If a tool helps you compare quality, behaviour, and likely cost on your own codebase, that is real value. A 2026 tool NEEDS to be focusing around models’ weaknesses, which starts with observability and monitoring. And, unfortunately, observability, control, and evaluation are still missing layers in many agent products. I highly recommend trying it out and taking the time to review agents’ code in general! I put the link in the comments if you want to try it. What do you care about more in coding agents today: more autonomy, or more observability?
Super excited to share that MLPerf Power (HPCA 2025) was selected for IEEE MICRO Top Picks 2025, 1 of the 12 most impactful computer architecture & systems papers of the year! Power consumption is the defining constraint for modern ML systems. Microsoft, Google, Amazon, Meta, and OpenAI have all announced plans for gigawatt-scale datacenters (for context, 5 GW = 5 nuclear reactors = Miami's power footprint). On the other end of the spectrum, we're anticipating billions of AI-enabled devices at the edge. We created MLPerf Power to be the industry-standard to measure, understand, and compare energy use across all deployment scales. We're excited to see that it's already impacting individual companies' strategies and has been incorporated into the IEEE semiconductor roadmap. We @MLCommons also collect and open source over 1,800 reproducible measurements from 60 diverse systems. These reveal several important insights that shed light on the nonlinear scaling of energy efficiency in modern systems and can enable many new data-driven optimization approaches. Just as @MLPerf aligned industry towards shared performance goals, we are hopeful that MLPerf Power will do the same for power and energy efficiency!
The LangSmith Signal: Azure's share of OpenAI traffic grew nearly 4x in under 3 months. We're sharing how devs are building agents, by the numbers. While most orgs started by connecting directly to OpenAI, over the past 10 weeks we've watched Azure's share of that traffic grow from 8% to 29%. We've analyzed this trend via LangSmith Observability data across more than 6.7 billion agent runs. Our hypothesis: 💡 Early adopters moved fast and went direct, but the enterprise wave is now arriving in force 💡 Azure gives teams the compliance, security, and procurement infrastructure they already have in place 💡 Azure traffic 4x-ing in 10 weeks likely indicates AI development is maturing quickly
