AI coding stack consolidates

AI coding products are starting to collapse into one software stack: the chat window, the editor, the terminal, and the agent are becoming the same workplace. (thenewstack.io) That shift is visible in product design. Cursor said last week that Cursor 3 is “a unified workspace for building software with agents,” replacing a single chat pane with an Agents Window built to manage multiple coding agents in one interface. (cursor.com) Anthropic is pushing the same direction from the terminal side. Its Claude Code product says it can read a codebase, make changes across files, run tests, and deliver committed code, and Anthropic said in December 2025 that Claude Code had been generally available since May 2025. (anthropic.com 1) (anthropic.com 2) OpenAI is stitching chat and coding together through pricing as well as product. On April 9, 2026, the company introduced a new $100-a-month ChatGPT Pro tier with 5 times more Codex usage than Plus, aimed at people running longer coding sessions. (community.openai.com) (help.openai.com) The pricing move landed as OpenAI’s coding product got bigger. CNBC reported on April 9 that Chief Executive Officer Sam Altman said Codex had 3 million weekly users, and The New Stack reported the user base had grown 5 times in three months. (cnbc.com) (thenewstack.io) The underlying technology is also converging. Anthropic’s Claude Opus 4.6 and Sonnet 4.6 both emphasize longer-running “agentic” coding tasks, while OpenAI’s pricing pages now describe Codex limits in terms of local messages, cloud tasks, and code reviews rather than simple chat turns. (anthropic.com 1) (anthropic.com 2) (developers.openai.com) That leaves less room for the old category labels. Cursor started as an artificial-intelligence-first editor, Claude Code as a terminal agent, and Codex as OpenAI’s coding assistant, but current coverage from The Verge and The New Stack describes them competing for the same developer workflow. (theverge.com) (thenewstack.io) The market pressure is now showing up in procurement, not just product demos. Anthropic already sells Claude Code with Team and Enterprise subscriptions, and OpenAI is using a middle $100 tier to narrow the gap between a $20 casual plan and a $200 heavy-use plan. (anthropic.com) (community.openai.com) Developers still see real differences in where the work starts — editor, terminal, or chat app — but the destination is increasingly the same: one place to assign coding work, watch agents run it, and decide what ships. (wired.com) (thenewstack.io)