Maintenance release on the extended-stable line — targeted security and reliability repairs, no new features. Notable for plugin authors: before_agent_start, root openclaw/plugin-sdk imports, providerAuthEnvVars, and channelEnvVars are scheduled for removal. OpenCode Go users get the corrected hy3 model identifier in place of the failing hy3-preview alias.
A 169-PR release dominated by security and reliability repair. Gateway HTTP now rejects disallowed browser origins before unauthenticated handling, external MCP loopback clients get short-lived session-bound grants instead of inheriting mutable child-process authority, Codex app-server commands require a real human or plugin approval, and secrets are redacted from MCP status output and Telegram-adjacent logs. Operators running exposed gateways should treat this as an upgrade release.
Fix-only release. Two matter for operators: Codex/GPT turns no longer stop mid-turn after a delivered progress message, and Memory Core startup conflicts no longer trap the Gateway in a fatal restart loop.
Large beta covering crash recovery for persisted state, durable message delivery across nine chat platforms, and conversation rewind/fork across web and native apps. Adds Claude Opus 5, Kimi K3, and GPT Live realtime support — but realtime Talk now requires a supported OpenAI Platform API key, as the Codex OAuth fallback is rejected.
Large beta covering persistence hardening (quarantine store, crash-recoverable SQLite snapshots, rejection of data-losing schema upgrades), durable message delivery across ten chat channels, conversation rewind/fork with branch switching, and a ticketed MCP Apps host with dashboards. Adds Claude Opus 5, Kimi K3, and GPT Live realtime — realtime Talk now requires a supported OpenAI Platform API key, as the Codex OAuth fallback is rejected. Also adds meeting guests for Teams/Zoom/Meet, a Wear OS companion, and in-process llama.cpp inference.
Large beta covering new model support (Claude Opus 5, Kimi K3, GPT Live via Codex OAuth), a ticketed MCP Apps host with dashboards, conversation rewind/fork with branch switching, durable message delivery across nine chat channels, in-process llama.cpp local inference, and a Wear OS companion. Affects anyone running OpenClaw as a gateway — state-durability and channel-delivery work changes recovery behaviour after crashes and restarts.
Prerelease carrying the 2026.7.2 cycle changelog: Control UI sessions on cloud workers, Codex/Claude sessions opened in terminals on their owning host, and a fix so channel allowlists no longer grant owner access. Fresh setups now default to GPT-5.6 (Sol), and Slack lifecycle reaction updates require opting in via messages.statusReactions.enabled: true.
Second prerelease of the 2026.7.2 cycle, publishing the same cycle changelog as later betas: remote coding sessions, headless-node capabilities, channel-allowlist owner-access fix, GPT-5.6 defaults for fresh setups, and Linux deb/AppImage packaging. Per-beta deltas are not stated in the source.
Large beta. Control UI sessions can run on cloud workers, and Codex, Claude, OpenCode and Pi sessions can be opened or resumed in terminals on their owning hosts. Also adds Linux deb/AppImage bundles, mobile Automations, and several authorization fixes — notably channel allowlists no longer grant owner access.
Stable 2026.7.1 release of the OpenClaw harness, drawing 3,063 contributions from 532 contributors across 2,018 credited PRs. Headline work is a Control UI and onboarding rebuild, substantial iOS/Android/macOS app updates, broader model and provider coverage including GPT-5.6, Tencent Hy3 and Meta Muse Spark 1.1, and Gateways that stop restarting forever when they fail.
Sixth beta of OpenClaw 2026.7.1. Adds Featherless, Claude Sonnet 5, Mythos 5, Meta Muse Spark 1.1 and ClawRouter, makes GPT-5.6 the default for new setups, reworks the Control UI around sessions, and stops Gateways from restart-flapping after repeated unclean boots. Slack operators must now opt in to lifecycle reaction updates.