Plus: Why Google’s giving away the sandbox Anthropic and OpenAI use ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­    ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­  
View in browser
The New Stack
Weekly Update  |  Issue 546
Together with
Budibase Logo Celery

TNS essential reads

Anthropic launches Fable 5.2 with a price cut and stronger coding scores

Claude Fable 5.2 cuts token prices by 20% and leads Anthropic’s coding comparisons, while a new Fast mode promises up to 2.5 times faster token generation.

Kubernetes, mark II? Why Google’s giving away gVisor, a code sandbox used by Anthropic and OpenAI

The sandbox heads to the CNCF alongside Agent Substrate and OpenRL, as Google bets neutral governance can make its AI agent stack an industry standard.

The Thoughtworks Technology Radar

The Thoughtworks Technology Radar isn't like other technology trend reports. It cuts through the hype and identifies only what deserves your attention. Read it now.
(Sponsored)

Cognition turned an open model into a coding beast. But it won’t release the weights.
Cognition’s SWE-2 Flash is a cheaper, faster coding model built on open-weight Inkling-Small — but the company is keeping its own weights private.

What five P99 CONF talks reveal about database trade-offs
Faster reads cost memory. Faster writes cost reads. Five P99 CONF talks show where engineers at ScyllaDB, AWS, and TigerBeetle chose to pay.

Busy GPUs can hide slow AI inference on Kubernetes
A healthy Kubernetes cluster and a busy GPU can still hide slow or incorrect AI answers. Profile inference before launch, then measure production against it.


Together with Budibase

Budibase 4.0: The Open-Source Platform for Enterprise Operations

Budibase transforms operations across your people, data, and systems. Build Agents, Apps & Automations with enterprise security and governance, and deploy on your own infrastructure.

  • Connect data, REST APIs, and LLMs

  • Set agent permissions & access

  • Enforce human escalations & approvals

  • Track every request & action
Find out more
Screenshot 2026-11-13 at 6.31.42 PM

AWS’s Hydro team wants the compiler to catch coding agents’ mistakes

Coding agents now write most of Hydro, AWS’s open-source Rust framework for distributed systems, and their pull requests keep getting bigger, with one reaching about 5,000 lines. In this episode, Hydro research lead Shadaj Laddad explains why code review and agent-written tests can’t keep up, and why the answer is correctness checks outside the agent’s “zone of trust,” enforced by the compiler. He also shares how the team is turning recurring design problems like idempotency into guardrails agents can share.

Catch the episode

Upcoming events & webinars

Nov 19
VIRTUAL
160 admin hours a year: how AI-guided diagnostics finds root causes faster
A file transfer fails at 2 a.m. Somewhere in a pile of logs, configs, and partner connections is the reason, and finding it can take hours. Across a year, MFT admins can lose up to 160 of them. Join us live on November 19 and watch AI work through a real failure scenario.
Register Now →
Dec 2
Virtual
Teaching AI agents to search like experts
AI agents are only as good as what they retrieve. On the BrowseComp-Plus benchmark, GPT-4.1 answered 93% of questions correctly when handed the right documents, but just 15% when it had to find them with keyword search. Join us live on December 2 to see what agents need from search, and how to give it to them.
Save your spot →
Dec 16
VIRTUAL
AI Sped Up Your Coding — But It Didn’t Speed Up Your Delivery
AI is helping developers write more code, but a Harvard study of 445 million work events found that most companies see only a modest bump in what actually ships. Learn why faster coding rarely means faster delivery, what the small group of high performers does differently, and how to find the real bottleneck between your code and your customers.
Register to join →
On demand
VIRTUAL
From Writing Tests to Verifying Outcomes: Who Tests the Code Your AI Agents Write?
AI assistants help teams ship code faster than ever, but test writing still moves at human speed, and every redesign breaks the suite. Watch on demand to see how agents can explore your app, generate tests from a plain-language prompt, and check outcomes instead of fixed steps, so tests keep passing when workflows change.
Watch on-demand →

From our sister site, roadmap.sh

Master SQL, the backbone of modern data

SQL remains the essential backbone of the modern tech stack, whether you're optimizing databases or scaling data-driven apps. Our sister site, roadmap.sh, offers a comprehensive SQL course built for hands-on, practical learning.

  • Go from foundational queries to advanced database management.
  • Learn through hands-on, practical exercises, not just theory.
  • Build skills that apply directly to real-world data work.
Browse all projects

Connect with The New Stack

LinkedIn
X
Facebook
Bluesky
Youtube

The New Stack is a media platform for the people who build and manage software the world relies on.

Sponsor this newsletter

The New Stack

1111 6th Ave Ste 550, PMB 50938, San Diego, CA 92101-5211

 

Not loving everything we send? You can update your preferences or unsubscribe from all email communications.