OpenAI AI Updates: August 26, 2026

1. OpenAI Shares First Results From Its Custom Jalapeño Inference Chip

OpenAI. OpenAI published first performance results for Jalapeño, a custom silicon chip built for AI inference that the company says delivers higher throughput, lower latency, and better power efficiency for modern models. The disclosure marks a significant step in OpenAI’s move toward in-house hardware to control cost and speed as inference demand scales. Owning the inference stack could reduce OpenAI’s reliance on third-party accelerators and lower the cost of serving frontier models. Source

2. Codex Scheduled Tasks Gain Event Triggers From Gmail, Slack, and GitHub

OpenAI. OpenAI added event triggers so ChatGPT scheduled tasks run in response to supported app activity instead of polling, including Gmail messages filtered by sender or subject, Slack channel activity, and GitHub pull request reviews, comments, commits, and merges. The feature is available on web and mobile for eligible plans. It moves Codex toward event-driven automation across common developer and workplace tools. Source

3. Codex Expands Browser Support and Adds WebMCP Site Tools

OpenAI. OpenAI extended Codex browser support to Edge, Brave, Opera, and Vivaldi, and added WebMCP site tools that let ChatGPT Work and Codex interact with functionality exposed by websites. The update also introduces cloud browser sign-in for eligible plans across web, iOS, and Android. The changes broaden where and how Codex agents can operate inside the browser. Source

4. OpenAI Introduces an Admin Plugin for ChatGPT Work and Codex

OpenAI. OpenAI launched an Admin plugin for ChatGPT Work and Codex that lets administrators analyze workspace usage, manage members and permissions, adjust limits, and act on admin requests from within the assistant. The tool brings enterprise administration directly into the ChatGPT interface rather than a separate console. It targets teams that need to govern larger deployments of ChatGPT Work and Codex. Source

5. OpenAI Disrupts a Covert Russia-Origin Influence Campaign

OpenAI. OpenAI banned Russia-origin accounts that used its models to promote a fake Israel-based think tank and a “sovereignty” index that praised Russia and criticized the West. The company detailed the operation as part of its ongoing threat reporting on malicious uses of its models. The disclosure highlights how state-linked actors continue to attempt AI-assisted influence operations and how OpenAI monitors and removes them. Source