OpenClaw has shipped v2026.4.11 as a stable release, promoting a packed beta into general availability less than a day after v2026.4.11-beta.1 appeared. That makes this morning's story pretty clear: the project did not just test a few ideas in beta, it moved quickly enough on confidence to bless the whole batch for broader use.
The release bundles a lot of surface area, but the strongest themes are easier plugin setup, a cleaner browser chat experience, richer memory observability, and a long tail of practical fixes for OAuth, transcription, sessions, and generated media.
v2026.4.11 Takes Yesterday's Beta Stable
The official v2026.4.11 release notes mirror the major changes that first showed up in Friday's beta. That matters because it signals the OpenClaw team was comfortable moving those changes from preview to stable almost immediately.
For operators, this is the safer framing: if you skipped the beta, you are not missing an experimental branch anymore. The new work is now part of the mainline release train.
The biggest additions include:
- better Dreaming / memory-wiki visibility for imported chats and compiled wiki pages
- a more structured Control UI / webchat renderer for assistant media, reply, and voice directives
- new plugin manifest setup descriptors for activation, auth, pairing, and configuration flows
- broader video_generate plumbing for URL-only delivery, provider options, role hints, and adaptive aspect ratios
- channel and platform improvements across Feishu, Microsoft Teams, WhatsApp, Telegram, and more
That is a lot for a single point release, but there is a coherent product story underneath it.
Plugin Setup Gets More Declarative
One of the most important changes is #64780, which lets plugin manifests declare activation and setup descriptors.
This is the kind of infrastructure work that often looks small in release notes and big a month later. Instead of hardcoding special-case setup logic into core flows, OpenClaw can now let plugins describe the auth, pairing, and configuration steps they need. That should make plugin onboarding less brittle and a lot more scalable.
If OpenClaw wants a healthier ecosystem around bundled and third-party plugins, this is exactly the sort of move you want to see. It shifts setup from tribal knowledge and ad hoc UI logic toward portable metadata.
Browser Chat Keeps Feeling More Like a Real Product
The web chat and Control UI changes from #64104 are another standout.
OpenClaw now renders assistant media, reply, and voice directives as structured chat bubbles, and it adds a new [embed ...] rich output tag while gating external embed URLs behind configuration. In practice, that means fewer raw protocol-like artifacts leaking into user-visible chat and cleaner handling for mixed text, media, and tool-driven responses.
That polish matters. OpenClaw has a lot of power-user depth, but browser chat is often where new users decide whether the system feels mature. Structured rendering makes conversations easier to read and makes tool output look intentional instead of improvised.
Memory and Dreaming Keep Growing Up
The Dreaming and memory-wiki work from #64505 also made the stable cut.
This release adds new Imported Insights and Memory Palace diary subtabs so users can inspect imported ChatGPT source chats, compiled wiki pages, and full source pages directly from the UI. OpenClaw's memory story has been getting deeper for weeks, and this release keeps pushing on the observability side of that vision.
That is important because memory systems get more trustworthy when users can inspect what was ingested and why it matters. Better visibility makes Dreaming feel less magical and more reviewable, which is exactly the right direction for long-term memory features.
The Fixes Are Not Filler
The stable release also lands several fixes that are easy to underrate but immediately useful:
- #64713 fixes Codex OAuth sign-ins that were failing with
invalid_scope - #64766 restores audio transcription for OpenAI-compatible multipart requests without weakening broader hostname validation
- #64723 fixes Google Veo requests by dropping an unsupported
numberOfVideosfield - #64918 preserves real inbound image paths for WhatsApp image edit flows
- #64869 keeps Telegram topic sessions on their canonical transcript path
- #64964 fixes MiniMax portal config patching so Bearer auth routing keeps working after re-auth
Put differently, this is not a release padded by one headline feature and a pile of trivia. There are real quality-of-life repairs here for people using media, sessions, OAuth, and multi-provider setups every day.
Why This Release Matters
OpenClaw v2026.4.11 feels like a maturity release.
The project is still adding big capabilities, but this build is more about connective tissue. Plugins get cleaner setup metadata. Browser chat gets better rendering. Dreaming gets better inspection surfaces. Media and auth paths get less fragile. Those are the changes that make a platform easier to operate, extend, and trust.
If you were waiting for the beta dust to settle, this is the green light. The work is now stable, and it touches enough core surfaces to make v2026.4.11 one of the more meaningful OpenClaw point releases of the month.
Read the full stable changelog on GitHub.