Visual Studio Code 1.104 brings major AI and safety improvements. GitHub Copilot now automatically selects the best AI model for each task, supporting GPT-5, Claude Sonnet 4, and Gemini Pro 2.5.
Forward-looking: OpenAI has officially launched GPT-5, more than two years after GPT-4 and a long parade of incremental upgrades. The company is pitching it as a "unified" AI that can seamlessly decide when to be fast, when to think deeply, and when to conserve resources. It's the kind of framing that suggests a new era for ChatGPT's 700 million weekly users and the broader AI market. But behind the marketing, GPT-5 is as much a strategic play in an intensifying AI arms race as it is a leap in capability.
Visual Studio Code has introduced a revamped tool picker and support for over 128 tools per agent request in MCP. Chat now features GPT-5 integration and checkpoints to restore previous states. Users can now manage multiple Git branches simultaneously, too.
Editor's take: Despite being hailed as the next step in the evolution of artificial intelligence, large language models are no smarter than a piece of rotten wood. Every now and then, some odd experiment or test reminds everyone that so-called "intelligent" AI doesn't actually exist if you're living outside a tech company's quarterly report.