OpenAI AI Updates: July 10, 2026

1. OpenAI Releases GPT-5.6

OpenAI. OpenAI released GPT-5.6, its new frontier model, positioning it as delivering more intelligence per token and stronger performance per dollar than its predecessors. The company describes the model as scaling capability on demand, allocating more compute to harder problems. It becomes the latest flagship in the GPT-5 series. Source

2. OpenAI Introduces ChatGPT Work

OpenAI. OpenAI introduced ChatGPT Work, an agent designed to take action across a user’s connected apps and files and to stay with a project for hours at a time. The company frames it as a partner for extended, complex work rather than single-turn queries. Source

3. OpenAI Makes GPT-5.6 the Preferred Model in Microsoft 365 Copilot

OpenAI. OpenAI announced that GPT-5.6 is now the preferred model powering Microsoft 365 Copilot across Word, Excel, PowerPoint, Chat, and Cowork. The company says the model brings stronger capabilities to those productivity surfaces. Source

4. OpenAI Launches the GPT-5.5 Bio Bug Bounty

OpenAI. OpenAI launched a GPT-5.5 Bio Bug Bounty, inviting external researchers to probe the model for weaknesses in its biological-risk safeguards. The program is part of the company’s approach to stress-testing safety mitigations before and after deployment. Source

5. Codex Joins the ChatGPT Desktop App

OpenAI. OpenAI added Codex to the ChatGPT desktop app for macOS and Windows, letting users edit Markdown and code directly, review GitHub pull requests, and work across repositories from within the app. The integration brings Codex’s coding workflows into the main desktop client. Source

6. OpenAI Ships Codex CLI Releases 0.143.0 and 0.144.0

OpenAI. OpenAI published back-to-back Codex CLI releases, with 0.143.0 enabling remote plugins by default with system proxy support and adding Amazon Bedrock GPT-5.6 models, and 0.144.0 adding usage-limit credits, new app-approval modes, and MCP authentication improvements. A follow-up 0.144.1 patch fixed standalone installation failures and code mode fallback reliability. Source