OpenClaw 4.11 Update: What's New (Release Notes Breakdown)

0:00OpenClaw 4.11 just dropped and here is everything you need to know. This isn't just a simple update with a few minor bug fixes. They shipped heavily in this update and we're going to be covering all in this video cuz there's a lot. Here's the one thing I think about in this update that I know is going to change a lot of people and about how they think about when switching to OpenClaw because here's the biggest update in quite a while. So, make sure you stick around because I'm going to be covering everything in this release. I want to start with the big one. I don't want to wait till the end for that one. ChatGPT import. That is huge. It's going to help bring a lot of users to the platform. Essentially, you can now import your entire ChatGPT history into

0:45your OpenClaw. That's the headline feature in this update and here's why I'm leading with it. If you've been using ChatGPT for the past two, three, four years, you have an archive of conversations sitting there, prompts you figured out, workflows you've iterated on, projects you've even researched, questions about your business, your niche, your audience, years of context about how you think and what matters and all of it is just sitting there in ChatGPT not doing anything, not feeding your AI setup. It's just archived sitting there. But with 4.11, you can now import that entire history into OpenClaw. Once you do, it shows up in the dreaming tab under two subtabs called imported insights and memory palace. And what that means is your

1:31agent gets to learn from all of those conversations automatically. It becomes context your agent can pull from when you're working together. So, think about what you actually want your agent to know about you. Your writing style, the way you structure problems, the industries you've researched, the workflows you've already figured out. That's all sitting in your ChatGPT history. You already did that work. Now, OpenClaw can use it. I think this is going to be really a bigger deal for two groups of people specifically. First, the people who've been on the fence about OpenClaw because they've got years of ChatGPT history and moving over feels like starting over. That objection is gone now. You can now bring your history with you. Real quick, before we keep going, if you're watching this and you want to actually build with some of these tools, you're going to want to

2:16check out our community down below, Shipping School. We have a full Claude code course, a full OpenClaw course, and four live bootcamps every single week where we actually help you get set up from scratch. Like actually set this thing up, not just watch a tutorial and figure it out by yourself. And we also provide one-on-one coaching so you could book a call with me, we could share screens, and I can help you get Claude code or OpenClaw running on your machine. That's it, no fluff. I built this community because watching YouTube's only gets you so far. We launched it just three days ago and we have over 55 members. You need people around you who are actually building, people who hold you accountable, and coaches who can help

3:01you when you get stuck. I'll put the link in the description down below. Get in now before the price goes up. The second type of people are the people who are already on OpenClaw, but their agent still feels kind of generic. Like it doesn't quite know who you are yet. This is a shortcut to giving your agent a much richer foundation. You know, one of the things I talk about inside of our community all the time is that the biggest value in AI agents isn't the AI itself. It's the memory, the context, the accumulated knowledge of who you are and what you're trying to accomplish. And the ChatGPT import is essentially giving OpenClaw a head start on building that for you, which I think is the highlight feature of this. The practical setup is pretty straightforward. You export your ChatGPT conversation data through ChatGPT settings. OpenClaw reads

3:48that from export file. I'm going to walk through this more in detail on one of our calls in our community soon because I want to make sure people know how to verify that the context is actually being used and not just sitting there as dead data. There's a right way to do it and I want to show you that and show the members as well. So, let's talk about the fixes in this update because there are some really important ones that I know a lot of people have been waiting on. The first one is the Codex OAuth. If you've tried to connect Codex to OpenClaw in the past few weeks and you hit an invalid scope error, this fix is the one you've been waiting for. What was happening is there was a URL rewrite in the OAuth flow that was breaking the authorization codes. So, the connection would start, it'd get partway through,

4:35and then it would fail. Depressing. It would look like an auth problem, but it was actually a routing issue on the backend. And that's been fixed in this latest update. And if you've hit that wall before and gave up on it, go and try again now. The second fix is audio transcription. This one's pretty significant. OpenAI transcription, Grok transcription, and Mistral transcription were all broken. Not degraded, not slow, but actually broke. The fix involved disabling pinned DNS for multipart requests specifically, which sounds technical, but the result is simple. Transcription works again. If you have voice workflows, Whisper pipelines, anything that relies on transcription, update to 4.11 and test it right away. That's a workflow blocker that you depend on and a lot of people depend on

5:21it. So do I. The third one is macOS talk mode. This one is going to matter to basically everyone watching this because you're probably on a Mac, I'm guessing. So, here's what was happening. You'd grant the mic permissions in macOS and then talk mode still wouldn't start. You'd have to go back and toggle it a second time to actually get it going, which is super annoying when you're in the middle of something and that's been fixed. Now, after you grant mic permission, it continues automatically. No second toggle needed. Small, minor thing, but you'll notice it immediately. The fourth thing is TTS in web chat. If you use audio replies in the OpenClaw web interface, the audio was getting attached to the wrong message in your chat history. So, you'd have a conversation, you'd go back to look at it later, and the audio responses were

6:06showing up in the wrong places. That's all being cleaned up. Audio replies now stay attached to the right messages. Your conversation history makes sense when you review it. The fifth thing is WhatsApp default accounts. If you had a named default WhatsApp account set up already, it was registering as a generic default instead of using the name you set. It's a small thing, but it matters if you've got multiple accounts configured. That's all fixed in the latest update. The new features beyond the ChatGPT import is pretty solid and it's a really nice list. Rich chat bubbles, the control UE, which is the web interface you use to interact with your OpenClaw agent, now renders responses as structured bubbles. So, media, voice directives, replies, all of

6:51it gets displayed in a cleaner, more organized way. They also added an embed tag that lets your agent output rich content directly to the response. If you spend time in the web UE, your agent's responses are just going to look a lot better, more readable, more organized, less wall of text, which is pretty cool if you're demoing this to someone or using the web UE as your primary interface. Video generation got a solid set of upgrades as well. Asset delivery is now URL-based instead of loading files directly into memory. So, if you're generating videos with large asset files, it's not going to clog your memory in the same way it did. You can now also pass reference audio inputs, which now opens up some interesting options for video generation workflows. Adaptive aspect ratio got added as well.

7:39So, video providers can handle different orientations without you needing to hardcode dimensions into your prompt every time. If you're doing automated video generation through OpenClaw, this is meaningful improvement in how much you can do without extra configuration. I run all my media through what is called kai.ai. It's just one API key and I can access multiple image-to-video models and I'm excited to see them shipping this within OpenClaw because I'm ready to move off of Kai and do all things internally. I do feel like it's not quite there yet for my use case, but I can see that they're improving with each update. Microsoft Teams now has full reaction support. If you've got Teams connected as a channel, your agent can now send and receive reactions

8:24properly using graph pagination and delegated OAuth to handle it correctly. If you're running a business environment on Teams, this makes the agent feel like a more native participant in the conversation instead of just a bot that sends text. Ollama users are going to notice faster model refreshes. Context window data and compatibility, metadata for Ollama models is now cached during the model discovery. Before this, every time OpenClaw needed to check what a model could handle, it would go fetch that information fresh. Now, it caches. Faster discovery, less unnecessary overhead. If you've got running local models through Ollama, you'll feel this especially during the setup when switching between models. Plugin setup descriptors. This is one of those infrastructure improvements that doesn't

9:09grab the headlines, but makes the platform more solid and reliable. Plugin manifests can now declare their own activation and setup flows. So, when you install a plugin, it can walk you through the authentication, pairing, and configuration steps in a structured way. Better onboarding for new plugins, less guesswork when you're adding something new to your setup. That's huge. And there is a new parity report model comparative. It compares ChatGPT 5.4 to Opus 4.6 across shared scenarios with stricter evidence criteria. So, if you're trying to figure out which model to run for which task, this gives you structured benchmark data to work from. If you're struggling to keep up with content, well, I'm about to save you about 40 days [music] worth of work. I built something called Content Machine.

9:55It's 10 AI agents that run on the Open Claw orchestration, and they handle everything. Scripts, thumbnails, X [music] posts, blogs, outreach, clips, newsletters, all of it. So, I went from 1,000 subscribers to 4,000 subscribers on YouTube in 7 days using this exact system. Every single morning I wake [music] up and the content's already done. I spend maybe 15-20 minutes reviewing and approving them, and I move on with my day. It works for any niche, fitness, [music] finance, real estate, marketing, whatever you are building, and it is 100% completely customizable to your use case. >> [music] >> So, you get the mission control dashboard, all of the cron jobs, everything I've built over the last 40

10:40days [music] helping me gain more and more people to subscribe and join the community. So, you plug in your own thing, and it molds it to you, it learns how you talk, and it writes so it doesn't sound like AI slop. $97 one time, it's not a subscription. I'll put the link down below, and you'll thank me later. You know, I've been running GPT 54 on most of my cron agents because it's just cheaper and performs well for those tasks. And then I use Opus 46 as my main session and my coordination agent, where I need the full reasoning capability. Having a formal parity report means you can make those model calls based on data instead of just a gut feeling. So, let me bring this back to what actually matters for your setup and what you should do right now. I tried to always tell you in this video that if there's a

11:26feature in this update, update it if you like the feature. 99% of the comments on these videos are I upgraded and my whole system broke. I want you to be sure that there's a feature in one of these updates that makes it certain that you want to update. But if you have everything working right now, you could skip past this update and just wait for the next one. So, that's what I want to say is like don't be in a rush to You should, if you are a ChatGPT user and you have 4 years of history, I think that's super helpful. But if you have things dialed in and you're working and you're selling agents as a service and you have your pipeline running, do you need an update? I don't know. You need to ask yourself that question. So, let me bring this back to what actually matters for your setup and what you should do right now. If you're brand new to Open Claw and you've got ChatGPT history, you need to start with that

12:12import. Do that before anything else. Giving more context to your agent is going to help tremendously. It's going to know who you are, your mission statement, other projects you've worked on. More context. Give your agent that foundation. It's going to make the whole experience feel more personalized way faster than if you start from scratch and try to build context manually. If you've been running Open Claw and you've been hitting the Codex OAuth error, update and try connecting Codex again. That's all resolved in this 411 update. If you have voice workflows or transcription in your pipeline, update it right now because transcription was broken before 411. It's all fixed. Don't keep running a broken transcription workflow. If you're on a Mac and you've been frustrated by the talk mode double toggle, that's gone. And if you're running a llama locally or doing video

12:57generation at any real scale, the performance improvements and video upgrades in this release are going to make a difference and you'll notice right away. How to update. If you've installed Open Claw through NPM, which is how most people do, just NPM update G Open Claw. That gets you to 411. If you're on a hosted setup, check your provider has already pushed the update. Also, another way to do it is just tell your Open Claw to update itself. Don't overcomplicate it. Full release notes are linked in the description down below so you can actually see the detailed release and go through every change in detail. For Shipping School members, I'm going to go through the ChatGPT import in a later boot camp call on why this makes a big difference for users that are big ChatGPT users. And I just want to make sure you know the impact of this and how

13:43to structure things in the dreaming tab to get the most value out of it. So, if you're not in the community, I'll put the link down below. We do nine calls a week, Q&A. We're basically really focusing on building AI agent and workflows in the community. 218 community members that have joined, and I launched the community 26 days ago. So, go update 411 if it's going to be a huge impact for you. I think it is, especially if you relied on ChatGPT a lot in the past. You know, it's one of those releases like even if you only care about one thing, that one thing is worth doing the update for. The ChatGPT import alone changes the whole conversation for people who've been thinking about making the switch. The fixes are things that were generally blocking real workflows. This one does matter. If you haven't subscribed to the

14:29channel already, please hit the subscribe button and hit the notification bell. I release videos four to six times a day, and if you want to keep up to date with AI breaking news and Open Claw use cases and Open Claw update videos, hit subscribe and we'll see in the next one. Be blessed.