OpenAI on Thursday launched GPT-5.4, the next version of its frontier model, calling it the “most capable and efficient frontier model for professional work,” reports TNS’s Frederic Lardinois in today’s top story.

The AI giant also noted in its announcement that GPT-5.4 combines the coding capabilities of its recently launched GPT-5.3-Codex model with improved support for spreadsheets, documents, and presentations.

The company argues that the new model is better at searching the web and maintaining context for queries that need longer thinking times. In addition, OpenAI improved the model’s computer-use capabilities and now allows it to more efficiently select the right tools when operating in an ecosystem with quite a few available.

It’s also the company’s most factual model yet, OpenAI claims, “with responses 18% less likely to contain errors, and individual claims 33% less likely to be false compared to GPT-5.2.”

Go deeper: OpenAI launches GPT-5.4 Thinking and Pro

More links for your Friday:

As workloads move on-premises, organizations are turning to Kubernetes-based internal platforms to deliver a public cloud-like experience. Don’t miss this live session on how to automate full-stack Kubernetes deployments with a single Git commit.

Claude Code is redefining the terminal, but moving from your first command to a professional production workflow requires a solid blueprint. Our sister site, roadmap.sh, just launched the Claude Code Roadmap to help you bridge that gap. This new roadmap covers everything from CLI fundamentals and MCP connections to advanced refactoring and CI/CD integration.