This week: GPT-5.4 benchmarks, Anthropic's first real measure of AI job exposure, agent orchestration frameworks, and why 2.5 million users quit ChatGPT.

The New Stack is a publication covering trends and technologies in cloud-native development, DevOps, and software delivery. Developers can learn about containerization, Kubernetes, and cloud computing, as well as explore topics such as microservices architecture, serverless computing, and continuous integration/continuous delivery (CI/CD) pipelines.

The New Stack

A weekly AI roundup covering GPT-5.4's launch with 83% GDPval benchmark scores, Anthropic's labor market study showing AI exposure is highest for computer programmers but hasn't caused broad unemployment, Google's agent-first CLI for Workspace, new open-source agent orchestration frameworks (Symphony and Paperclip), and the fallout from the Pentagon's supply chain risk designation of Anthropic. The controversy drove 2.5 million users to quit ChatGPT, pushed Claude to #1 on the App Store, and helped Anthropic reach a $19 billion annualized revenue run rate.

OpenAI GPT-5.4 launches, AI gets its own jobs report, Anthropic surges after U.S. ban

OpenAI launched GPT-5.4, and Codex hit 1.6 million users

Agent Orchestration is becoming a critical layer

OpenAI is in damage control. Anthropic is breaking records.