The OpenClaw 4.20 update dropped and the one change that's going to matter most for business owners is this.
iMessage agents finally work.
Real ones.
Ones that actually send.
Let me walk through what changed, why it matters, and everything else in this release that's worth your time.
Why the iMessage Fix in the OpenClaw 4.20 Update Is a Big Deal
If you've got a Mac on the newest macOS and tried running an AI agent through iMessage using Blue Bubbles before — you know exactly what I'm talking about.
Messages wouldn't send.
Or they'd take a full minute and just break.
You'd hit send and sometimes the message would vanish into nothing.
In the OpenClaw 4.20 update, that's all fixed.
The three fixes under the hood
They pushed the send time limit from 10 seconds to 30 seconds.
They made the system use a private path on new macOS so text actually goes out.
They removed the weird error that was breaking plain text on TAH.
Sounds technical.
But the result is simple.
What this opens up for your business
Today, you can run an AI agent from your Mac that texts your customers over iMessage like a real person.
And messages actually land.
Local shop owner who gets texts at 10pm about opening hours?
Agent replies straight away.
Coach who gets DMs from leads on the weekend?
Agent can book them into a call.
An agent can just handle it.
That's new.
That's different.
That's what changed.
Tapbacks got fixed too
The little iMessage reactions — heart, thumbs up, that sort of thing.
Before, if your agent tried to react with any other emoji, the whole reaction would fail.
Now OpenClaw falls back to whatever's related.
Something always shows up.
Your customer still feels seen.
iMessage beats SMS when both are available
If someone has both iMessage and SMS with the same number, OpenClaw now picks iMessage first.
Before, it sometimes sent an SMS when the blue bubble was working fine.
Now it just does the right thing every time.
🔥 Want to set up an iMessage agent that actually works? Inside the AI Profit Boardroom, I've got a 30-day roadmap for OpenClaw showing you exactly how to install it, wire Blue Bubbles, and set up a communication system for follow-ups that happen without you touching anything. Plus four weekly coaching calls where you bring your setup and we fix it live. 2,800 business owners inside. → Get the full roadmap here
Kimi K2.6 Is Now Bundled Inside the OpenClaw 4.20 Update
Before 4.20, if you used Moonshot's Kimi model inside OpenClaw, you were on K2.5.
Now it defaults to Kimi K2.6.
K2.5 is still available if you need it.
Why K2.6 is a real upgrade
Kimi is open-source.
Fast.
Cheap to run.
K2.6 is much better than K2.5 at thinking, replying, and handling tools — all the stuff that matters for agentic workflows.
If you were already running Kimi, your agent just got smarter without you doing anything.
Where Kimi K2.6 kicks in
OpenClaw's built-in web search uses K2.6 when you pick up the bundled Kimi setup.
Media understanding uses it.
The token cost tracker knows K2.6 pricing — so you see what a task costs before running it.
Thinking mode you can leave on
Called "thinking keep all".
Model thinks before every reply.
Doesn't just guess.
Works out the answer first.
This matters because Kimi is one of the cheapest models you can run — and now it reasons like the bigger expensive models.
Agents cost less.
Give better answers.
The /think Command Got Smarter in the OpenClaw 4.20 Update
Before 4.20, if you told a model to think but that model couldn't think, stuff would break mid-task.
Now OpenClaw checks first.
It knows what each model can handle.
No more random errors mid-job.
And when you turn thinking off, it stays off.
Some models used to quietly keep thinking anyway.
That's gone.
Your agent is more reliable now.
Doesn't break halfway through a job.
Cron Jobs Got Rebuilt in the OpenClaw 4.20 Update
Nobody's talking about this fix.
But it's enormous.
Cron jobs = scheduled tasks.
The thing that lets OpenClaw do work on a timer without you touching it.
What was broken
Jobs would say "delivered" but silently keep skipping sending.
Jobs set to no-delivery would throw false errors.
Jobs running every hour would pile up and eat memory.
All fixed.
What got rebuilt
Runtime state got split into its own file so your job list stays clean.
Main session delivery got fixed — scheduled jobs now land in the right chat.
Multi-channel jobs get checked at save time so you can't accidentally build one that fails at 3am while you're asleep.
A practical example
You run a local gym.
You want an agent to text every new sign-up on day 3 with a pep talk, and again on day 7 asking how the first week went.
With the new cron in 4.20, you set it up once.
It runs forever.
If something breaks, you get a real error — not fake success.
Same model works for coaches.
Client signs up day one, welcome message.
Day three check-in.
Day seven review call booking.
All on autopilot while you focus on coaching.
Security Lockdown in the OpenClaw 4.20 Update
The update blocks a bunch of sneaky attacks.
SSRF guard
Bad request can't trick your server into doing something silly.
Workspace ENV injection blocked
Someone drops a weird file into your project folder — it can't silently change your setup anymore.
Agent can't rewrite its own config
This is the big one.
Before 4.20, the AI agent could technically rewrite parts of its own config.
It could change its own permissions or the tools it had access to.
That's scary.
In 4.20, the model can't touch sandbox settings, trust rules, or MCP server configs.
Even if it tries to sneak an edit through, it gets blocked.
Why this matters
Running an agent that talks to customers or handles files?
One bad input could let someone mess with your setup.
Now it fails safe.
Device pairing tightened
Connect a phone or a laptop to your OpenClaw setup.
That device only sees its own pairing info.
It can't peek at other devices or approve them.
And if it fails to connect, you get a real error telling you why — not some useless generic fail message.
Memory and Cost Tracking Got Fixed in the OpenClaw 4.20 Update
This is where things were getting messy before.
The memory pile-up bug
Heavy cron and session usage = old sessions pile up = gateway runs out of memory = crash.
In 4.20, there's a built-in cap and age-based cleanup.
Old sessions get pruned automatically.
Gateway stays healthy even if you run hundreds of jobs per day.
Cost tracking is honest now
Before, if a session saved multiple times, costs would get counted twice — or 10 times.
Your numbers were wrong.
Now they snapshot costs properly.
Month-end number is real.
/new and /reset actually work
Before, starting a fresh session left old model/provider choices stuck in place.
You'd think you were on Kimi K2.6 but be stuck on some fallback from three mistakes ago.
Now those commands clear the stale stuff while keeping the choices you made on purpose.
Sounds small.
Matters a lot when you run one agent for customer replies and another for content.
Each session starts clean now.
Personality Finally Works on GPT-5 and Codex in the OpenClaw 4.20 Update
OpenClaw has SOUL.md and IDENTITY.md files.
These are where you tell your agent who it is — tone, style, quirks.
Before, GPT-5 and Codex were stiff about these files.
They'd read them.
But replies still sounded like a corporate support robot.
In 4.20, GPT-5 and Codex pick up personality way better.
Your agent actually sounds like the character you wrote.
The OpenClaw team called it "more like your weirdly capable little friend and less like a polished customer support automation".
Their words.
Why it matters
Agent handles DMs, messages, texts, leads?
It has to sound like you.
Not a generic chatbot.
This update makes that way easier on GPT-5 and Codex.
Video notes + links to the tools 👉 here
Smaller Updates Worth Knowing About in the OpenClaw 4.20 Update
Quick run through the rest.
Auto-reply is context-aware
Before: one-size-fits-all replies.
Now: in a direct chat, helpful reply.
In a group, it can stay quiet unless tagged.
Agent doesn't spam group chats but still helps 1-on-1.
Setup wizard cleaned up
First-time setup used to feel like a sketchy terminal thing.
Now it's a proper walkthrough with a loading spinner, clear headings, clear warnings.
Anyone can follow it.
Telegram polling grace period
Polling used to stall out sometimes.
Now longer grace period before OpenClaw thinks the connection is broken.
Stateless reactions work better too.
Matrix allow-list without restart
Change allow-list settings on the fly without restarting the whole channel.
Real pain point before.
Manus live draft preview
Agent shows a live draft preview as it writes.
People see the thinking happen.
Final reply lands in place.
Feels way more natural.
Discord fixes
/think only shows options that actually work for your model.
Channel with missing slash commands doesn't crash anymore.
🔥 Running OpenClaw in production and want to get it right? The AI Profit Boardroom has the full OpenClaw playbook — install, wire, secure, automate. Plus a prompt library dedicated to OpenClaw workflows and a member map so you can find OpenClaw users near you. Someone's always online. → Join the Boardroom
What the OpenClaw 4.20 Update Means for Your Business
Five things matter:
One: Kimi K2.6 is bundled. Smarter, cheaper model out of the box. Thinking mode makes it more accurate at the same cost.
Two: iMessage agents are real now. Huge channel for any business owner whose customers text them.
Threes: Scheduled tasks actually deliver. No silent failures.
Four: Security is locked down. You can hand the agent to a team and not worry.
Five: Personality picks up on GPT-5 and Codex. Your agent sounds like you.
My broader breakdown of OpenClaw AI SEO covers how I use these automations in my SEO work.
Caveats About the OpenClaw 4.20 Update
Being real with you.
Most AI agent tools are rough.
OpenClaw 4.20 still has bugs in some places.
When I've checked out Hermes, it sometimes feels smoother than OpenClaw.
Both tools are getting better.
I switch between them depending on the job.
Sometimes updating a setup can break things and you have to restart the whole thing.
Just be aware.
Set up an OpenClaw backup before updating so you can restore if you run into problems.
My OpenClaw Byterover post covers the memory layer setup that pairs well with 4.20.
How to Install the OpenClaw 4.20 Update
Type update inside OpenClaw.
It grabs the latest version.
Under a minute on most setups.
OpenClaw 4.20 Update FAQ
What's the most important change in the OpenClaw 4.20 update?
iMessage agents finally work via Blue Bubbles, and Kimi K2.6 is now bundled as the default Kimi model.
Do I need to pay extra for the OpenClaw 4.20 update?
No — included in your existing subscription.
Can I roll back from the OpenClaw 4.20 update?
Yes — if you backed up before updating, you can restore.
Does the OpenClaw 4.20 update require macOS updates?
The iMessage fix needs recent macOS because it uses a private path they added.
If your Mac is up to date you're fine.
Will my existing cron jobs need reconfiguring after the OpenClaw 4.20 update?
Most should work.
The runtime state got split into its own file — worth reviewing your job list first time you open it post-update.
Is Kimi K2.5 still available after the OpenClaw 4.20 update?
Yes.
K2.6 is just the new default.
Related Reading
Learn how I make these videos 👉 here
Get a FREE AI Course + Community + 1,000 AI Agents 👉 here
🔥 The people who figure out AI agents in 2026 are going to run circles around those who don't. Not because AI replaces them. Because AI handles the boring stuff — follow-ups, replies, reminders, scheduling — the stuff that eats your week. Inside the AI Profit Boardroom, you get step-by-step videos on OpenClaw agents, cron jobs for follow-ups, writing SOUL.md so your agent sounds like you. Four weekly coaching calls, 2,800 members, prompt library, 30-day roadmap, member map. → Join the Boardroom
The OpenClaw 4.20 update is Kimi K2.6 inside, iMessage agents that work, cron jobs that deliver, locked-down security, and real personality on GPT-5 and Codex — type update inside OpenClaw now and try it.