Broadcasting from
the core of tech_
KernelCast decodes AI, agents, infrastructure, and devtools — short, sharp videos for people who build. Every drop ships with a written companion post: the take, the sources, the tools.
CPU vs GPU vs TPU vs NPU vs LPU: What's the Difference?
Five chips, five names, five companies — and when it comes to AI, all of them are doing the same math. What actually differs is how each one does it. A plain-English tour of the CPU, GPU, TPU, NPU, and LPU, and the one idea that makes all five make sense.
read the take→
Fable 5 vs GPT-5.5 vs Opus 4.8 vs GLM-5.2: How the Four Actually Differ
Four flagship models, one honest question: how do Fable 5, GPT-5.5, Opus 4.8, and GLM-5.2 really differ — in character, in output, and in cost? A decision guide, with real-world coding tests as the evidence.
read the take→
Top 5 AI Customer Support Platforms of 2026 (Ranked & Compared)
Choosing the right AI customer support platform isn't just about rankings—it's about finding the one that fits your business. In this video, we compare the top 5 AI customer support platforms of 2026, highlighting their strengths, use cases, and what makes each one stand out. Watch the full ranking to see which platform takes the #1 spot.
read the take→
Top 5 AI Coding Platforms Every Developer Should Use in 2026
We scored the five biggest AI coding platforms — Claude Code, Cursor, Codex, Copilot, and Replit — across five dimensions and ranked them #5 to #1. The twist: the whole field fits inside 1.1 points. There's no single winner, just a winner for how you build.
read the take→
Sakana Fugu: AI Model That Matches Fable 5 on Key Benchmarks
Sakana's Fugu isn't a bigger model — it's a conductor. One endpoint that answers easy questions itself and hands hard ones to a team of specialists. Fugu Ultra hits 73.7 on SWE-Bench Pro, beating GPT-5.5, Opus, and Gemini — but it still rents its brains.
read the take→
Claude Fable 5: The Model Too Dangerous to Ship — Until Now
Anthropic just shipped Claude Fable 5 — the first public model in a new Mythos tier above Opus. It found 271 zero-days in Firefox, tops every coding benchmark, and costs 2× Opus. Here's what actually matters.
read the take→
OpenClaw vs Hermes Agent: Which Should You Pick?
OpenClaw and Hermes Agent are 2026's top personal-AI-assistant platforms — both well-funded, both NVIDIA partners, built on completely different stacks. We score them across five technical dimensions and call an honest verdict.
read the take→Five ways we cover the stack.
The whole idea in a hundred seconds. No filler, no intro music.
> filter by 100sThe deep dive — architecture, tradeoffs, and why it actually matters.
> filter by explainedWe order the field and defend every placement.
> filter by rankedTwo contenders, one rubric, one verdict.
> filter by vsA new release, reacted to while it's still warm.
> filter by dropThe full archiveevery transmission→Get the drop in your inbox.
New video, the written take, and the links — delivered when it ships. No spam, no noise.