OpenAI launches ChatGPT Work, an enterprise agent merging Codex into the desktop app

Alongside the GPT-5.6 models, OpenAI unveiled ChatGPT Work — a distinct product that consolidates ChatGPT and the Codex coding app into a single agentic experience. Greg Brockman framed it on X as 'an agent for your most ambitious work,' emphasizing that users can invoke it from mobile or web in addition to desktop, 'no need to leave your laptop cracked open.' The agent is designed to handle complex, multi-step workflows that span multiple applications, positioning it directly against Anthropic's Claude Code and the wave of agentic coding tools shipping this week.

The product is powered by the new GPT-5.6 family, with Sol handling long-horizon agentic tasks. By merging Codex into a unified desktop app, OpenAI signals a strategic bet that developers and knowledge workers want one integrated agent rather than separate chat and coding tools. This also tightens OpenAI's enterprise story just as Microsoft named GPT-5.6 the 'preferred model' for Copilot.

The competitive backdrop is intense. Meta's Muse Spark 1.1 launched the same day pitching agentic performance at a 'very low price,' and SpaceXAI's Grok 4.5 is now default in Grok Build and available in Cursor. François Chollet captured the mood: 'It's mind-blowing how fast agentic coding has progressed in the past 6 months. It's a completely different world now.' The open question for enterprises is reliability on real engineering tasks versus vendor benchmarks — a recurring theme in developer threads debating which agentic coding tool actually wins. ChatGPT Work's success will hinge on whether it delivers dependable multi-step execution in production, not just polished demos.