OpenAI has released GPT-5.6, a new model series with three variants (Luna, Terra, Sol) and expanded thinking levels, along with a unified ChatGPT Work app that merges the previous Codex and macOS ChatGPT applications.
Users should be aware that higher thinking levels consume usage limits much faster, and OpenAI temporarily removed usage caps during recent bug-fix work, creating the risk of exceeding weekly limits in a single session.
What happened
OpenAI released GPT-5.6 models to all users, featuring three variants (Luna, Terra, Sol) each with five thinking levels (light, medium, high, xhigh, max) plus a new Ultra mode. The macOS ChatGPT app and Codex app have merged into a single unified application called ChatGPT Work, and a new ChatGPT Sites plugin enables building hosted websites with optional login integration.
Why it matters
The merge simplifies the user experience by combining coding and general-purpose tools into one app, while the expanded thinking-level options let users trade off speed and accuracy per task. However, higher thinking levels consume usage limits significantly faster—OpenAI even temporarily removed the 5-hour usage limit during weekend bug fixes, meaning users can deplete weekly allowances in a single session.
What to watch
Sol excels at UI design and writing at Max thinking; Terra offers minor improvements over GPT-5.5 with better steerability; Luna works best for clearly-defined tasks but struggles with ambiguous prompts. The models' Computer Use feature (self-driving cursor control) is available now, particularly effective at Sol medium/high thinking levels.
Ask the AI about this article →
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.
For example, today's edition would include:
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. Q&As are published on this page for other readers too.
CBTS Technology Solutions LLC launched Forge Agents, a platform that turns a plain-language job description in…
Imec CEO Patrick Vandenameele said at SEMICON Taiwan 2026 that the Belgian research center is broadening its c…
Alphabet's AI Overviews now reach over 2.5 billion monthly users through Google Search, and its ad business ge…
Amazon Web Services (AWS) has integrated its fully managed data warehouse service, Amazon Redshift, with Agent…
Visual Studio Code 1.135 now includes an experimental 'Rubber Duck' feature that lets developers request a sec…
Sonos announced a new app update with generative AI features, a new soundbar called the Beam Ultra, and its se…
