OpenAI releases GPT-5.6 with three models, merges Codex into ChatGPT Work

OpenAI has released the GPT-5.6 model family to all users, bringing significant product and architectural changes. The release centers on three models—Luna, Terra, and Sol—each equipped with five thinking levels (light, medium, high, xhigh, max) plus a new Ultra mode that enables these models to deploy subagents for more complex reasoning. According to the article, Sol performs particularly well on UI work and improves markedly at Max thinking for writing tasks; Terra represents a minor step forward from GPT-5.5 with improvements in UI and writing skills and greater steerability; Luna operates effectively on well-defined tasks but can misinterpret ambiguous prompts.

The most visible product change is the merger of OpenAI's macOS ChatGPT app and Codex into a unified application called ChatGPT Work. Users update their Codex app to receive the new combined interface, which maintains separate modes fine-tuned for coding versus non-coding work. A new plugin called ChatGPT Sites allows users to build hosted websites with optional "Login with ChatGPT" integration. During the weekend integration work, OpenAI reset usage limits 4–5 times while fixing bugs introduced by the app merge and temporarily removed the 5-hour usage cap, creating a window where users could exhaust weekly allowances in a single session.

Usage consumption is a critical trade-off: the body notes that higher thinking levels deplete usage limits much faster than lower ones. One user reported nearly exhausting their Codex usage after experimenting with Ultra mode for the first time, leading them to adopt Luna at xhigh thinking for most building and creativity work, background agents for harder tasks, and Sol medium for day-to-day productivity. The GPT-5.6 models in the Codex app excel at Computer Use—the ability to self-drive a cursor to open applications, click buttons, and interact with screen elements—with Sol at medium or high thinking recommended for initial exploration of the feature.

OpenAI's GPT-5.6 release marks a significant consolidation of its product surface. By merging Codex and ChatGPT into a single ChatGPT Work application, the company is streamlining the user experience—though this integration required multiple usage resets during the weekend to fix bugs introduced by the app merge, and the temporary removal of usage limits suggests the transition was operationally complex. The introduction of three distinct models (Luna, Terra, Sol) with granular thinking-level control reflects a design philosophy that lets users optimize for their specific task and budget constraints rather than forcing all workloads through a single model. However, the body notes that higher thinking levels consume limits much faster, creating a practical tension: users who rely on Max or Ultra thinking may exhaust their allocations quickly, particularly if they default to expensive reasoning chains for routine tasks.

No comments yet. Be the first to share your thoughts!