OpenAI launched ChatGPT Work this week, a cloud-based AI agent that connects to email, Slack, Google Drive, Microsoft 365, and calendars to execute multi-step professional workflows autonomously. The product runs on GPT-5.6 Sol and represents OpenAI’s direct answer to Anthropic’s Claude Cowork, according to Forbes.
Alongside the launch, OpenAI discontinued its standalone AI browser, Atlas, less than a year after release, and merged the Codex coding agent into a revamped ChatGPT desktop application.
The Desktop Consolidation
The Codex merger creates what OpenAI internally calls its “superapp”: a single desktop interface with built-in browser capabilities and computer control. Applications Chief Fidji Simo drove the consolidation, which The Wall Street Journal reported was prompted by product fragmentation that “had been slowing [the company] down and making it harder to hit the quality bar [they] want.”
OpenAI also introduced two new reasoning modes alongside GPT-5.6. Max mode allocates additional compute for complex problems, while Ultra coordinates four agents in parallel for demanding workflows, according to Forbes.
GPT-5.6 Performance and Pricing
GPT-5.6 Sol comes within one point of Anthropic’s Claude Fable 5 on the Artificial Analysis Intelligence Index (59 vs 59.9) while costing roughly one-third as much per task, according to Forbes. Pricing is set at $5/$30 per million input/output tokens for Sol, $2.50/$15 for Terra, and $1/$6 for Luna.
Sol leads the Coding Agent Index at 80 points, outperforming Fable 5 on software engineering benchmarks including Terminal-Bench 2.1 (91.9%).
Atlas Shutdown
The Atlas browser shutdown after less than a year signals a broader industry pattern: dedicated AI browsers are losing to integrated agent interfaces. Browsing capabilities now live inside the ChatGPT desktop app rather than a standalone product. Gadgets Now reported the discontinuation alongside the product consolidation.
Government Oversight Delay
The public launch was delayed after the Trump administration requested early access over cybersecurity concerns. OpenAI had to comply with a “limited preview” for government-approved partners before receiving approval for wider release. OpenAI stated in a blog post that it does not believe “this kind of government access process should become the long-term default,” but called it “the strongest path to broader availability in the coming weeks.”
The Competitive Landscape
ChatGPT Work puts OpenAI in direct competition with Anthropic’s Claude Cowork for enterprise agent workflows. VentureBeat noted that OpenAI positioned the product as a more affordable option, leaning on GPT-5.6’s cost advantage. In early internal testing, OpenAI reported that nearly 100% of internal teams now use ChatGPT Work, with finance teams reducing month-end close from days to hours.
The consolidation strategy is clear: one app, one interface, agents that do everything. The question is whether unified platforms win over enterprises already invested in specialized tooling.