HeadsUpAI

OpenClaw Adds Grok OAuth and Realtime Voice to Self Hosted Assistant

OpenClaw, an open-source self-hosted AI assistant that runs across messaging platforms, released version 2026.5.18 with native xAI Grok OAuth support. This allows subscribers to authenticate securely without manually managing API keys. The update also introduces a realtime Android Talk Mode featuring streaming microphone input and tool-result bridging.

This release continues a broader effort to modularize the OpenClaw runtime for production-grade reliability. By ensuring generated media stays within Telegram forum topics and making browser automation aware of modal dialogs, the platform reduces the friction that typically disrupts autonomous agent workflows in complex, multi-channel environments.

You can now deploy new skills for Python debugging and meme generation while benefiting from faster gateway startup times. The update is available for self-hosted instances, with the Mac app receiving a settings redesign. These refinements follow a security roadmap aimed at hardening the platform's agent execution environment.

View the full update on github.com
OpenClaw🦞
OpenClaw🦞
@openclaw
X

OpenClaw 2026.5.18 is live 🤖 xAI/Grok OAuth + sidecar auth fixes 🎙️ Realtime Android Talk Mode 💬 Telegram media + forum-topic delivery fixes 🪟 Browser dialogs visible + answerable A week of polish, plumbing, and fewer papercuts. https://t.co/pwD67BdnhK

97retweets887likes
View on X

Still wondering? A few quick answers below.

OpenClaw is an open-source, self-hosted AI assistant designed to run on a user's own hardware. It functions as a personal agent that can manage emails, calendars, and home automation through messaging apps like Telegram and Discord. The platform uses a plugin-based system to add specific skills and supports various large language models.

The v2026.5.18 update introduces OAuth support for xAI Grok, which is a secure authorization standard that lets applications access data without sharing passwords. This allows subscribers to connect their accounts and use Grok models within OpenClaw directly through a login flow rather than manually managing and inputting sensitive API keys.

The Android application's Talk Mode has been upgraded to a realtime voice session using a gateway relay. This system supports streaming microphone input and provides realtime audio playback with tool-result bridging. Users can now see on-screen transcripts of the conversation while the assistant processes and executes commands through voice interaction.

OpenClaw has improved how its agents interact with web browsers by making them aware of modal dialogs, which are pop-up windows that block main page interactions. The system now reports when an action is blocked by a dialog and allows the agent to see and answer specific pending dialogs using a dedicated command.

OpenClaw performed a reliability pass on its Telegram integration to ensure that generated media and replies are delivered correctly to forum topics. By preserving topic identifiers across agent handoffs, the system prevents messages from leaking into the base chat and ensures that replies maintain the correct thread IDs within organized group topics.

Share this update