OpenAI has announced the general availability of its GPT-5.6 models, which it has claimed offer best-in-class coding, knowledge work, and cybersecurity performance alongside more affordable AI outputs.
The GPT-5.6 family encompasses three models: Sol, Terra, and Luna.
Sol is OpenAI’s new frontier offering, which OpenAI said sets a new bar for AI performance at a lower cost than rival offerings. Terra is aimed at offering more balanced performance for day-to-day tasks, while Luna is intended to offer lower cost and latency.
The independent AI analysis firm Artificial Analysis scores GPT-5.6 Sol as the new leader in all its coding benchmarks, outperforming models such as Anthropic’s Claude Fable 5, xAI’s Grok 4.5, and Google’s Gemini 3.1 Pro, with GPT-5.6 Terra taking the second spot.
For example, Sol achieved a record 88 per cent in Terminal-Bench v2, a benchmark that evaluates models on 89 real-world terminal tasks, while Terra achieved 84. Fable 5, by comparison, achieved 82.
It added that in tests, GPT-5.6 Sol proved competitive with Claude Fable 5 across a wide range of tasks at approximately a third of the cost.
OpenAI initially staggered the model’s launch, making it accessible only to pre-approved organisations, in line with a request by the Trump administration.
GPT-5.6 Luna is available for a fifth of Sol’s price, with performance as good or better as GPT-5.5 and Claude Opus 4.8 according to independent benchmarks.
Alongside the release OpenAI announced that GPT-5.6 is now the preferred model in Microsoft 365 Copilot, acting as the backbone for all tasks completed with the built-in AI tool. Users can still choose alternative models depending on their task.
Nitin Agrawal, president for Copilot & Agents Core at Microsoft, said: “Using Copilot powered by OpenAI’s latest model, customers will be able to produce more polished outputs in Word, Excel, PowerPoint, Cowork, and Copilot Chat, whether they are drafting documents, analyzing data, creating presentations, or collaborating across teams.”
The model has already been praised by OpenAI’s enterprise partners.
Arjun Sambamoorthy, VP and CTO of AI software and platform at Cisco, said: “GPT‑5.6 consistently stays focused through long-running tasks, makes excellent use of tools, and gets to high-quality solutions with little steering. For research and design work, it produces clear reports and intuitive diagrams that help our teams understand complex systems and move faster.”
