OpenAI’s launch of ChatGPT Work and GPT-5.6 Sol has sparked major backlash, with users reporting rapidly drained usage limits, a confusing desktop redesign, and broken multi-agent workflows. Thibault Sottiaux admitted the company “didn’t get everything quite right” after a wave of complaints. Key issues include easy access to high-cost compute settings that burned through budgets faster, a messy desktop overhaul that hid chats and projects, and unclear messaging about Codex—leading users to think the coding tool was being killed off. OpenAI has since reset usage limits twice in one day, adjusted default settings, and promised a bigger fix next week that restores the sidebar and improves usage metrics. The company also clarified Codex isn’t going anywhere, despite its desktop app initially greeting users with “Codex is now the ChatGPT app.” Why it matters: This botched rollout shows that even as AI product offerings explode, user experience and clear communication can make or break adoption—especially for tools developers rely on for daily workflows.