Microsoft announced on July 9 that GPT-5.6 is now rolling out to Copilot experiences across Word, Excel, PowerPoint, Copilot Chat, and of course Copilot Cowork.
Cowork users also get to choose between GPT5.6 Sol and Terra models options.
GPT‑5.6 is OpenAI’s latest flagship model series, which delivers more useful work from every token (and Copilot Credit), with stronger performance and on demand capability for the most complex tasks.
In their blog, Microsoft say that
Copilot may automatically use GPT-5.6 when it’s best suited for the task at hand, but where model selection is available for users (such as in a Chat and Cowork), will be able to choose GPT-5.6 directly from the model selector.
GPT 5.6 Sol is the flagship GPT‑5.6 model. It introduces max reasoning effort (letting the model think longer on hard problems), which uses multiple sub‑agents to complete complex workflows requiring planning, iteration, and tool use. Sol leads major benchmarks such as Terminal‑Bench 2.1 (up to 91.9%) and delivers top‑tier cybersecurity performance while using far fewer tokens than competing models. It also includes OpenAI’s strongest safety framework to date, with enhanced protections against misuse. Sol is designed for advanced coding, biology, cybersecurity, and long‑horizon reasoning tasks
GPT 5.6 Terra is the balanced, everyday GPT‑5.6 model,offering GPT‑5.5‑level capability at roughly half the price, making it ideal for general knowledge work, business tasks, and daily productivity scenarios. Terra is designed to be cost‑efficient while still supporting strong reasoning, coding, and workflow automation.
Open AI said in their blog that
“GPT‑5.6 with our most robust safeguards to date, designed to be resilient against determined and adaptive misuse without broadly limiting legitimate work…. During the preview, we worked closely with expert organizations and with trusted partners to pressure-test defenses and strengthen safeguards before broader launch. The resulting system layers protections trained into the model with real-time checks, monitoring, and access calibrated to trust and risk.”
Leave a Reply