OpenAI has begun a phased rollout of its GPT-5.6 artificial intelligence model, structured into three tiers—Luna, Terra, and Sol—to comply with requests from the U.S. government. The release aims to balance technical performance with regulatory alignment, positioning the model as a highly efficient tool for professional workflows while competing directly with Anthropic’s Claude Fable 5.
GPT-5.6 Performance and Tiered Architecture
The GPT-5.6 rollout introduces a tiered system designed to address specific user needs. According to OpenAI, the Luna model serves as the base, optimized for cost-efficiency. The Terra model acts as the mid-range option for daily tasks, while the Sol model offers the highest performance. The Sol variant includes a new “Ultra” mode, which manages multiple agents in parallel to resolve complex tasks.
Technical benchmarks highlight significant shifts in efficiency. In the “Agents’ Last Exam,” which evaluates long-horizon workflows across 55 fields, the GPT-5.6 Sol model scored 53.6%, outperforming the Claude Fable 5 “adaptive reasoning” mode by 13.1%. Furthermore, in the “Artificial Analysis Coding Agent Index,” the Sol model reached a record 80%—surpassing Fable 5 by 2.8%—while using less than half the output tokens, effectively reducing costs by roughly one-third.
The GPT-5.6 Sol model achieved a new record in coding tasks while simultaneously cutting token consumption by more than half compared to rival benchmarks.
Integration with ChatGPT Work
Alongside the new models, OpenAI launched ChatGPT Work. This tool integrates with professional applications including Slack, Notion, Microsoft 365, and Google Drive. It functions by synthesizing disorganized data into polished, editable documents, spreadsheets, or presentations. Currently available for macOS and Windows, a web-based version is expected to follow shortly.
Market Competition and Industry Reactions
The release occurs amid intense competition between OpenAI and Anthropic. Industry feedback has been mixed, reflecting different user priorities. Pietro Schirano, CEO of MagicPath AI, stated on X that GPT-5.6 is the “best model I’ve ever used,” citing its speed and design capabilities. Conversely, investor Matt Shumer noted that the Claude Fable 5 outperformed GPT-5.6 in several of his internal tests.
Dan Shipper, CEO of Every, provided a comparative framework for these tools: “If you need to cross the galaxy, use the Fable. If you need to get around town with the best tool available for the purpose, use the 5.6.” While Anthropic’s models are often cited for raw intelligence in specific scenarios, the industry consensus frames the GPT-5.6 as a reliable, fast, and cost-effective solution for daily professional routines.
Anthropic’s Response with Claude Cowork
Anthropic has countered this release by expanding its autonomous agent, Claude Cowork. Previously restricted to computers, Cowork now functions via the cloud for Max plan subscribers. This allows the agent to execute tasks—such as report structuring and email analysis—in the background, even when the user is offline. The new “Dispatch” feature also enables remote management of these active sessions via mobile devices.
Frequently Asked Questions
What are the three versions of GPT-5.6?
OpenAI released the model in three tiers: Luna (base, cost-efficient), Terra (mid-range, balanced), and Sol (high-performance, including the “Ultra” agent-coordination mode).
How does GPT-5.6 compare to Claude Fable 5?
Data from the “Artificial Analysis Coding Agent Index” shows GPT-5.6 Sol outperforming Claude Fable 5 in coding tasks by 2.8% while using significantly fewer tokens. However, some users still report that Fable 5 holds an edge in raw intelligence for specific, complex scenarios.
What is ChatGPT Work?
ChatGPT Work is a desktop application for macOS and Windows that connects to professional tools like Slack and Google Drive to automate the creation of documents, spreadsheets, and presentations from scattered data.
Stay updated on the latest developments in AI and enterprise technology. Subscribe to our newsletter to receive daily briefings on the tools shaping the future of work.
