Anthropic Just Doubled Your Limits (Do This Now)

0:00Anthropic just did something that's going to save you a ton of money if you use Claude code and it expires in just 8 days, so you need to hear this right now. Anthropic, the company behind Claude, announced that they're doubling your usage limits during off-peak hours. This is a crazy time to be alive because if you are worried about cutting costs, now is your time to actually build stuff if it's off-peak hours. So, this isn't just like a little bump in credits and usage. This is literally double the usage and it applies to Claude code, which is the tool I use every single day to build up my entire operation. So, if you're a builder and if you're using Claude for coding or writing or anything, this is

0:45crazy. Here's how it works. Off-peak hours are technically 2:00 p.m. to 8:00 a.m. Eastern time. So, if you're on the West Coast like me, that's 11:00 a.m. to 5:00 a.m. Basically, the majority of my working day falls into off-peak hours, which means I'm getting double the usage for the same price and it runs until March 27th. So, you still have 8 days or so to take advantage of this. And this is the crazy part and where it gets more interesting is the Reddit community on the Claude subreddit figured out something that Anthropic probably didn't intend as the main takeaway. They said to add ultra think to every single prompt during off-peak hours. This gives your prompt more deeper reasoning when

1:30you're talking to Opus or Sonnet and it is just more precise and it understands the context better. You could use think, think harder, or ultra think. And if you didn't know that, that is a huge hack to getting quality output. So, what does this mean exactly? You know, Claude has different thinking modes. You know, there's regular thinking like I said, there's extended thinking, and then the ultra think. They actually took it away for a few months and I was kind of depressed. So, ultra think is the deepest level where Claude really takes the time to reason through the complex problems. Normally, ultra think burns through your usage limits faster because it uses more computing power, but if you're getting double the limits during off-peak hours, you can use ultra think on everything and still come out ahead. This is crazy. For me personally, I'm

2:16loving it. Let me tell you why. I run Opus 4.6, which is Anthropic's most powerful model for my main agent on my content pipeline. Every morning, all these agents fire off. We got 13 in total now and, you know, we got my Intel agent that fires off at 5:00 a.m. pulling, scraping the entire internet for relevant stuff in the AI space. And this is all occurring off-peak. My X post agent in podcast outreach is Inspect the Deck. He's amazing. And my thumbnail agent is You God. He fires off at 6:30 off-peak. So, basically, my entire morning cron jobs in pipeline, the core of the content business, runs during the window where I now get double the usage. So, if you think about what that means, I can run more complex

3:01prompts. I can let my agents think deeper. I can use ultra think in my script generation and get better quality output without worrying about hitting limits halfway through the day. That's pretty cool. Real quick, before we keep going, if you're watching this and you want to actually build with some of these tools, not just watch videos about them, then you're going to want to check out our community down below, Shipping School. We have a full Claude code course, a full open Claude course, and [music] 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 open Claude running on

3:46your machine. That's it, no fluff. I built this community because watching YouTube's only gets you so far. We launched it just 3 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 you when you get stuck. I'll put the link in the description down below. Get in now before the price goes up. And let me back up for a second because some of you might not know what Claude code is and why it even matters so much. Claude code is a command-line tool. It's there from Anthropic and it lets you lose use Claude directly in your terminal. You know, you could give it access to your files, your project folder, your code base, basically your entire project. And it could read through your code. It could write code. It could run commands. It could debug

4:31issues and all from the terminal. It's basically like having a senior developer sitting right next to you, eating cheeseburgers, watching while you work. And they never get tired. They never get frustrated and can hold your entire project in its head, especially with the 1 million context window that they just came out with about a week ago. You know, I use Claude code through open Claude, which is the AI agent platform I run everything on, my entire businesses. And open Claude connects to Claude's API and uses Claude code under the hood for a lot of the heavy lifting. So, when Anthropic doubles my limits, it directly impacts how much my agents can do in a day. More limits means more agent runs, more scripts, more content, more everything. Now, let me break down why

5:16this even matters beyond just the limit increase. The move tells you something about Anthropic's strategy. They're not just giving you more usage out of the goodness of their hearts. There's more to it. They have compute capacity sitting idle during the off-peak hours. Their GPU clusters are humming along, maybe 40 to 50% of utilization, while everyone in the US is asleep. Instead of letting those GPUs sit there doing nothing, they're incentivizing you to shift your work to those hours. It's the same thing electric companies do with off-peak electricity rates. You use more during the cheap hours, save money, and the company balances their load. It's pretty smart. And this is brilliant for builders like us. If you can schedule your heavy compute work during the off-peak hours, you can effectively double your output for the same costs. This is a a great reason that you need

6:01to ship that idea, to build that idea. I know you have it in your mind. Now is the time to go, baby. I already do this naturally because my cron jobs run every morning, but if you're doing manual work with Claude code, think about shifting your hardest coding sessions to after 2:00 p.m. Eastern. That's basically the middle of the day. There's no excuse not to ship. And that's when you want to be asking Claude to like refactor your entire code base, help you plan your product requirements document, build a new feature from scratch. Save the simple questions for peak hours and when everyone else is hammering those servers. Let me give you my practical tips for maximizing this window. There are things that I'm actually doing right now and this is not hypothetical advice. First, you need to schedule your heavy agent during off-peak hours. So, if

6:46you're using open Claude or any agent framework, set your cron jobs to run between 2:00 p.m. and 8:00 a.m. Eastern. My entire pipeline already runs in this window by accident because I'm on the West Coast and my agents fire off between 5:00 a.m. and 7:00 a.m. Pacific. But if you're on Eastern time, you might want to push your morning jobs to a little bit later or run them in the evening instead. The second thing is you got to use ultra think for everything. During off-peak, you've got double the headroom, so you don't need to hold back on thinking depth. If you're writing a complex script or generating content, debugging a tricky issue, slap the ultra think on every single prompt. The quality difference between regular thinking and ultra think is actually pretty noticeable and you're going to go through a lot less headaches. It's like the difference between asking someone a

7:32quick question while they're walking by versus sitting them down and saying, "Take your time. Really think about this." And the output is just better. It's basically more thorough. It catches edge cases. It considers things in regular mode that regular mode would basically miss. The third thing is batch your coding work. If you've got a list of features to build or bugs to fix, don't spread them throughout the day. Stack them during the off-peak window. You have 8 days to do this. I've been doing this with my mission control dashboard. I save up all the feature requests and bug reports during the day and then hit them all during my evening coding sessions when I've got double the limits. It's basically like having happy hour for coding. Same price, double the drinks. The fourth thing I would do is this is the most one most important one is people will miss this is use the extended window to experiment. One of

8:19the biggest barriers to trying new things with AI is worrying about burning through your limits on something might not work. And I get it. We've all been there before. You want to try building something new, but you're like, "What if it eats all my tokens and I can't finish my actual work?" We've all been there. But with double the capacity, this is the time to try that wild idea you've been sitting on. Build that agent you've been thinking about. Google Stitch 2.0 just came out and you can literally take the MCP to connect into Claude code and have beautiful landing pages done in seconds. Do it. Test that workflow. Try a new skill on Claude Hub. Whatever. You've got the headroom right now. Use it. You have 8 days. The next thing, if you're on a team, to coordinate your schedules. If you've got maybe two or three people using Claude code, stagger your heavy sessions so you're all

9:04hitting the off-peak window, not all at the exact same time. That way you get the most value across the whole team without everyone competing for the same resources. Here's something I want to be clear and honest about. This promotion ends on March 27th. We don't know if Anthropic is going to extend it or make it permanent or they're going to hook you as a customer. My guess is they're testing this to see how users respond. If enough people shift their usage to off-peak and then the load balancing works, they might make it a permanent feature with like a tiered pricing model. But for now, treat it like it's ending on the 27th and maximize every single day you have left. And you know what? One of our members inside of Shipping School, Rusty, has been tracking his API spend religiously. He's

9:49spent about 100 bucks over 5 weeks on the Claude API usage. With double off-peak limits, someone like him could do the same amount of work for significantly less money or do twice the work for the same cost. Either way, it's a win and I think that's the kind of practical impact this has for real builders, not just people watching from the sidelines. Let me also put this into perspective and in context with everything else that's been happening with Claude Code this month because this past month they've been absolutely shipping everything. They release something every day at this point. They dropped voice mode for Claude Code, which lets you literally talk with Claude while it's coding for you. Just like speak your internet instructions out loud while you're looking at the code and they launched the forward slash

10:34loop command as well, which lets Claude run the iterative development cycles without you having to keep prompting it. They expand context window up to 1 million tokens, which means Claude can hold your entire code base in its head at once. A million tokens is like a thousand pages of code. They made Opus 4.6 the default model for paying users and now they're doubling the off-peak usage. Feels like Anthropic is going all in on developers right now. They literally love us. And as someone who builds with their tools every day, I'm here for it. Claude Code is printing money for them. They reportedly hit a two and a half billion dollar run rate this year and they're reinvesting that back into making the developer experience that much better. This is the kind of cycle you want to see as a user. The product gets better, more people use it, more revenue comes in, product gets

11:20even better. It's a great beautiful flywheel. One more thing I want to mention. If you are on the Claude Pro plan and you're thinking about whether it's worth upgrading, it's double off-peak usage makes the math even more compelling. You're essentially getting a Pro Plus experience during off-peak hours for the price of Pro. And if you combine that with 1 million token context and ultra think, the value per dollar right now is the best it's ever been for Claude. I'm not saying rush out and buy it. I'm saying if you're already on the fence, the math is in your favor. So to recap, Anthropic doubled Claude Code usage during off-peak hours. That's 2:00 p.m. to 8:00 a.m. Eastern. It runs until March 27th. Add ultra think to all of your prompts during those hours for better quality output. Schedule your heavy work in the off-peak window. Batch

12:06your coding sessions. Experiment with the stuff you've been putting off. And if you're running agents like I am, make sure your cron jobs fire during that window to maximize the value. And if you want to learn how to set up AI agents and AI agent pipeline and agentic workflows that takes advantage of stuff like this, that's what we cover inside the community. There's a reason why 106 people have joined in the last 10 days because we have six live bootcamps every single week to help grow your business, to help you learn how to use AI tools, and how to make money doing that and with them. So, click the link down below. I got it there and get in before the price goes up. We're also releasing four to six long-form videos every single day on this channel. So, if you just like AI stuff, make sure you subscribe and we'll see you in about an hour. Bye-bye.