OpenClaw released version 2026.7.1 on July 13, 2026, delivering one of the platform’s most substantial updates to date. The release redesigns the Control UI, simplifies installation and onboarding, improves the official desktop and mobile apps, expands AI model support and strengthens integrations with messaging platforms and external coding agents.
A More Powerful Control UI
The redesigned Control UI moves OpenClaw closer to a complete browser-based workspace for managing AI conversations, sessions and background tasks.
Users can now organize sessions into custom groups, pin important conversations, rename or archive sessions, mark messages as read or unread and open multiple sessions in resizable panes. These layouts can be preserved after refreshing the browser, making it easier to manage several agents or projects without opening multiple tabs.
The updated interface also includes a live Tasks page for monitoring and canceling authorized background work. Tool calls, progress updates, approvals, results and errors are displayed in cleaner expandable rows, reducing clutter in complex conversations.
Usage reporting has also been expanded. Administrators can review estimated costs, token consumption, provider usage, model activity, quotas, balances and budgets. When properly configured, the dashboard can also display OpenAI and Anthropic billing information alongside OpenClaw’s internal session analysis.
Easier Installation and Onboarding
OpenClaw v2026.7.1 significantly reduces the technical barriers involved in getting started.
New Mac users can open the application and begin using a local agent without installing Homebrew, opening Terminal or providing administrator access. The macOS onboarding process now tests the selected AI provider before the first conversation and offers clearer recovery options when authentication fails.
Android users receive a guided setup process covering QR pairing, Gateway approval, permissions, troubleshooting and device readiness. OpenClaw’s conversational Crestodian setup can also walk users through provider credentials, model selection, channel connections and configuration repairs.
Improved iOS, Android and macOS Apps
The official OpenClaw apps received major improvements to navigation, chat, voice, files, permissions, localization and Gateway recovery.
Conversations and recent transcripts can now remain available during temporary Gateway outages. Users can read cached sessions while offline, queue supported text messages and automatically send them in the proper conversation after reconnecting.
The iOS and Android apps preserve scroll positions as messages and tool activity arrive. Apple Watch users can dictate messages, hear final responses and control active requests directly from the watch.
The macOS app now includes a complete session sidebar, transcript exporting, message copying, slash-command completion, cost information and additional tools for managing conversation history.
Expanded AI Model and Provider Support
OpenClaw v2026.7.1 expands compatibility with hosted, managed and locally installed AI models.
Organizations with access to GPT-5.6 preview models can select Sol, Terra or Luna through supported OpenAI and Codex routes. The release also adds or improves support for Tencent Hunyuan Hy3, Meta Muse Spark 1.1, Claude Sonnet 5, Claude Mythos 5, LongCat-2.0, Gemini and local Ollama models running on paired computers.
Local Ollama support is especially useful for organizations that want selected tasks to run on their own macOS, Linux or Windows computers. Administrators can separately disable node-based inference when local model execution is not appropriate.
Stronger Codex and Coding-Agent Workflows
OpenClaw now works more effectively with coding agents operating outside the current conversation.
The new openclaw attach workflow can connect Claude Code to a selected Gateway session using temporary, revocable credentials. This avoids exposing broader process-level credentials and helps limit access to the intended session.
Codex sessions can resume previous work, delegate assignments to native subagents and return those results as tracked tasks. Important coding conversations can also be pinned, renamed, archived, restored and monitored as durable project threads.
Better Messaging Integrations
Telegram, Slack, Discord, WhatsApp and Apple Messages all receive reliability improvements.
Telegram can process complete photo albums instead of passing only one usable image to the model. Slack threads retain conversation history more consistently and are less likely to duplicate replies. Discord improves attachments, voice sessions, unread notifications and recovery after temporary Gateway interruptions.
WhatsApp now handles concurrent replies and reconnections more reliably, while Apple Messages improves attachment routing, typing indicators, replies and conversation continuity across remote Mac configurations.
Smarter Scheduled Work and Remote Browser Control
Scheduled OpenClaw jobs can now react when an external command finishes or a monitored condition changes. This allows the full agent to remain inactive until there is meaningful new information, reducing unnecessary processing.
Remote browser control has also been expanded. Operators can pair the OpenClaw Chrome extension with a remote Gateway, share selected authenticated browser tabs and revoke access by removing a tab from the OpenClaw group. Agents can wait for delayed browser downloads and save completed files to approved local locations.
Workspace Terminals Across Web and Mobile
Authorized administrators can now open protected workspace terminals through the Control UI, iOS app or Android app.
Browser-based terminals can be docked, resized and placed beside active conversations. Sessions can reconnect after page reloads, temporary network interruptions or device sleep while preserving recent output. Terminal access remains subject to Gateway ownership, authentication and security policies.
A Major Step Toward a Unified AI Workspace
OpenClaw v2026.7.1 is more than a collection of minor fixes. The release makes the platform easier to install, easier to navigate and considerably more capable across desktop, mobile, messaging and coding environments.
The most important shift is OpenClaw’s evolution from a conversational agent interface into a unified workspace where users can manage sessions, agents, scheduled jobs, coding tools, terminals, browser activity, model usage and connected devices from one environment.