OpenAI GPT-5.6 powers Microsoft 365 Copilot

// 1h agoPRODUCT UPDATE

OpenAI GPT-5.6 powers Microsoft 365 Copilot

OpenAI has announced that its new GPT-5.6 model family is now the preferred model powering Microsoft 365 Copilot, bringing advanced reasoning and improved efficiency to core office applications. The rollout is underway globally, enabling enhanced document drafting, data analysis, and presentation features across the productivity suite.

// ANALYSIS

Microsoft's immediate integration of GPT-5.6 highlights its commitment to maintaining the lead in enterprise AI productivity, even as it abstracts away the raw model names for the end-user.

* GPT-5.6 introduces specialized model variants—Sol (flagship), Terra (mid-range), and Luna (speed)—tailored for different tasks.

* The update brings tangible quality improvements to core knowledge-work activities like document synthesis and data visualization.

* This release further solidifies the strategic alliance between Microsoft and OpenAI in the enterprise market.

// TAGS

microsoft-365-copilotgpt-5.6openaigenerative-aioffice-productivityenterprise-ai

DISCOVERED

1h ago

2026-07-10

PUBLISHED

2h ago

2026-07-10

RELEVANCE

9/ 10

AUTHOR

sama

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

MODEL22m ago

GPT-5.6 excels as agentic orchestrator

AI researcher and educator Elvis Saravia shared his positive experience using OpenAI's newly released GPT-5.6 model, expressing surprise at how effectively it performs in high-level orchestrator roles, specifically noting its strength in verifying and advising within developer workflows.

LAUNCH29m ago

0xDesigner launches Codex Marketplace for coding agents

0xDesigner announced the Codex Marketplace, an ecosystem and hub for discovering, rating, and installing extension plugins, custom skills, and Model Context Protocol (MCP) connectors for the modern agentic coding ecosystem. By providing a unified directory, the marketplace streamlines developer workflows, allowing them to install specialized tools (such as UI design generators or API connectors) directly via the Codex CLI or desktop application, moving agent development away from complex prompt engineering toward modular plugin selection.

UPDATE33m ago

OpenAI clarifies GPT-5.6 reasoning effort levels

In response to developer questions about token consumption and model performance, OpenAI's Nik Pash clarified that reasoning effort levels (like "xhigh") are rough product labels rather than fixed, apples-to-apples token budgets across model versions. Because GPT-5.6 spans a wider capability range than GPT-5.5, the "xhigh" setting on GPT-5.6 is not equivalent to the same setting on GPT-5.5 and can consume significantly more tokens to execute deeper reasoning.