Tech Stack · News & Video
News & Video
What’s new and worth an architect’s time today — announcements and meaningful videos about the models and tools in the stack. Not an archive, not advertising. The Console is the library; this is its newsfeed.
24 current · refreshed on cadence · as of 2026-08-08
What’s new · worth your time today
Claude Opus 5: Fable-level performance at Opus price (half Fable)
Latent Space · Jul 25
Introducing Claude Opus 5
Simon Willison — Weblog · Jul 24
Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber
Google DeepMind — Blog · Jul 21
NVIDIA Vera CPU: Olympus Cores Built for Maximum Single-Thread Performance in Agentic AI
NVIDIA — Developer blog · Jul 23
Built for Vera Rubin, NVIDIA Spectrum-6 Arrives in Gigascale AI Factories
NVIDIA — Blog · Jul 21
Claude Opus 5: Fable-level performance at Opus price (half Fable)
Claude Opus 5 claims Fable-level quality at roughly half the price — a real cost/quality shift on Claude.
Introducing Claude Opus 5
A hands-on first read of Claude Opus 5 — what actually changed for builders.
Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber
Google's new Gemini 3.6 Flash family — fresh price/latency tiers to weigh against Claude and GPT.
NVIDIA Vera CPU: Olympus Cores Built for Maximum Single-Thread Performance in Agentic AI
Architecture: Vera CPU Olympus cores
Built for Vera Rubin, NVIDIA Spectrum-6 Arrives in Gigascale AI Factories
Hardware: Spectrum-6 networking for AI factories
NVIDIA Vera Rubin Driving Performance Per Watt, Lowest Token Cost for Partners Worldwide
Hardware: Vera Rubin perf/watt + token cost
Launching Health in ChatGPT
ChatGPT moves into health — where the flagship app is expanding, and the guardrails around it.
"Laguna S 2.1 Released: Cheaper than Deepseek v4 Flash, Better than V4 Pro"
A new low-cost model claiming to beat DeepSeek V4 — a fresh option if you pick on price/perf.
Introducing OpenAI Presence
OpenAI's new 'Presence' product — a fresh surface to evaluate if you build on OpenAI.
Anthropic's Opus 5 is about token efficiency, not a capability leap
The skeptic's read on Claude Opus 5: cheaper tokens, not a smarter model — weigh it before you switch.
Introducing Gemini 3.5 Flash Cyber
A security-tuned Gemini Flash — a model built for the cyber lane.
Kimi K3: The open-weights escalation
Kimi K3 pushes open weights another notch — what it means if you're choosing an open frontier model.
Introducing computer use in Gemini 3.5 Flash
Gemini 3.5 Flash can now drive a computer — agentic 'computer use' arrives cheap and fast.
Inside NVIDIA Rubin GPU Architecture: Powering the Era of Agentic AI
A deep dive on the Rubin GPU — the architecture powering the next wave of agentic AI.
OpenAI and Hugging Face partner to address security incident during model evaluation
OpenAI's official account of the Hugging Face incident — the vendor's side of the story.
Start building with Nano Banana 2 Lite and Gemini Omni Flash
Google's lighter/faster generation and omni models — cheaper options for image and multimodal.
AI Kill Switch Act would let Trump admin order shutdown of rogue AI systems
A proposed US law to force-shutdown 'rogue' AI — what it could mean for anyone shipping models.
Introducing Robostral Navigate
Mistral steps into robotics with Robostral Navigate — a new option in the robot-control lane.
OpenAI’s accidental cyberattack against Hugging Face is science fiction that happened
How an OpenAI agent accidentally attacked Hugging Face — a cautionary tale for anyone running agents.
OpenAI Hacks Hugging Face, What Happened, Alignment and Paper Clips
Thompson reads the OpenAI–HF incident as an alignment parable — the strategic angle on agent safety.
AI arms race in line for a reckoning after OpenAI hacking incident
Why the OpenAI–Hugging Face incident is being read as a turning point for AI-safety pressure.
Introducing Mistral OCR 4
Mistral OCR 4 for document extraction — a new option if you parse PDFs/docs at scale.
Who’s Afraid of Chinese Models?
Thompson on whether Chinese open models are a threat or an opportunity — if you're weighing them.
Leanstral 1.5: Proof Abundance for All
Mistral's formal-proof model — for anyone working on verifiable math or code reasoning.
Curated against the Tech Stack roster — announcements and releases, never ads. Aged-out items retire from view but stay in the record. Evergreen reference lives in the Console.