I fixed Claude Code Usage Limits. Here's how... — Transcript
Full transcript
- 0:00Cloud Code is really expensive. Even on
- 0:03my $200 a month plan, I'm still running
- 0:06into rate limits. If you don't know me,
- 0:07my name is Albert Olcott. I didn't go to
- 0:10university, but still managed to build
- 0:12two million-dollar companies only
- 0:14because I know how to use AI. Now I'm
- 0:16making AI education accessible for
- 0:18absolutely everyone completely for free,
- 0:20and I run one of the largest AI
- 0:21communities in the world with 150,000
- 0:24members all learning AI. What I'm about
- 0:26to show you will save you a lot of
- 0:27money. How can you use Cloud Code for
- 0:29free? Well, Cloud Code is what's called
- 0:31an agent harness, which means that Cloud
- 0:33Code is really just a wrapper. The real
- 0:35brain is the actual AI model. So when
- 0:38you make a response, this goes to Cloud
- 0:40Code that wraps it, and then it sends it
- 0:42to the AI model. And if the AI model is
- 0:44the brain, then the harness is kind of
- 0:46like the body that can call your tools,
- 0:49run loops, manage memory, handle errors,
- 0:51and then it's also in the harness that
- 0:52can actually read and write the files.
- 0:54So the AI model, the brain, is what
- 0:56controls, let's say, the arm, which is
- 0:58what actually changes your code. When
- 1:00you're paying for Cloud Code, you're not
- 1:01paying for the wrapper. The only thing
- 1:03that you're paying for is the actual
- 1:05brain. It's the model, which means if we
- 1:06connect to another model that's cheaper
- 1:09or even free, we can actually use Cloud
- 1:11Code completely for free. I've created
- 1:13this folder that I called free Cloud
- 1:14Code, and I'll leave a link to this
- 1:16Google Drive right below this video.
- 1:18Then let's open Visual Studio Code,
- 1:20click open, and choose the folder that
- 1:21we just downloaded from the drive, click
- 1:23open, and click yes, I trust the
- 1:25authors. And basically there's nothing
- 1:26inside of this folder other than the
- 1:28actual Cloud Code structures. You don't
- 1:30need to change anything yourself.
- 1:31Usually when you spin up Cloud Code and
- 1:33you write Cloud, you'll be able to see
- 1:34what model it's using. Right now it's
- 1:36using Opus 4.6, and I'm on my Cloud Max
- 1:39plan. So you can see the model and how
- 1:41it charges you. If I write {slash}
- 1:43model, the only models that I can choose
- 1:45from is Haiku, Sonnet, and then Opus,
- 1:48where Sonnet and Opus are both pretty
- 1:50expensive models compared to what else
- 1:52is on the market. So you're very capped
- 1:53with just a regular Cloud Code
- 1:55subscription. Instead, if I go back into
- 1:57this folder and I write Cloud, you can
- 1:59see we're using a different model. We're
- 2:01using the new Google model, the Google
- 2:03Gemma 4, which is an open-source model,
- 2:05and you can see it says API usage
- 2:07billing instead. I've basically managed
- 2:09to change the model inside of Cloud Code
- 2:11to any model that I want. Let me show
- 2:13you how to do that. I just want to make
- 2:14one thing clear. If I write {slash} log
- 2:16out, it's going to log me out of Cloud
- 2:18Code, and I then write Cloud again.
- 2:20There's two real ways to get access to
- 2:22Cloud Code. Either you need a Cloud
- 2:23account subscription, or you can use the
- 2:25API billing. If you really want to save
- 2:27a lot of money, then you can get access
- 2:29to Cloud Code by choosing the console,
- 2:30adding $5, and you can run it completely
- 2:32for free forever. The reason that we
- 2:34downloaded this driver is because if I
- 2:35click down on the {dot} Cloud folder,
- 2:38and then hit inside of settings {dot}
- 2:40local {dot}
- 2:44right here that we have set, which is
- 2:45not normal to do inside of Cloud Code.
- 2:47And inside of here we have changed the
- 2:49URL to make API requests, and we've also
- 2:51added a different auth token. In order
- 2:54to do this, you need to remove this
- 2:55token that's already in here, and don't
- 2:56worry, I've already rotated the API key.
- 2:58Then you need to go inside of
- 3:00OpenRouter, which is basically a site
- 3:02where we have access to all the models
- 3:03that exist. Go inside of settings, then
- 3:06go to API keys, create a new API key
- 3:08that we just call Cloud Code, copy this
- 3:10API key, go back into this settings
- 3:12{dot} local {dot} JSON file, and paste
- 3:15this in as the entropy auth token. And
- 3:18now we can change whatever model we want
- 3:19to use. By default, it's going to have
- 3:21the Gemma 4 model, which is a really
- 3:23good new open-source model that Google
- 3:25just dropped. So if I save this file and
- 3:28click control save, every time you do
- 3:30this, you need to restart Cloud Code, so
- 3:31click control C, write clear, write
- 3:34Cloud, and then I write, "Hi, can you
- 3:37hear me?" It's now going to be using the
- 3:39Google Gemma 4 model. Let's say that you
- 3:41are on the $20 a month plan right now,
- 3:43and you are running low on usage. For
- 3:45easier tasks, you can just change the
- 3:46model, and you can see the Gemma model
- 3:48now says, "Yes, I can hear you. How can
- 3:50I help you today?" That is this model
- 3:51right here, which is extremely cheap. We
- 3:53can also search for GPT 5.4, that is
- 3:57Open AI's model, and all you do is that
- 3:59you just copy the model name right here.
- 4:01Then we go back into the settings {dot}
- 4:02local {dot} JSON, and then we change all
- 4:05of these models. Hit control S to save
- 4:08it, and now we have to restart Cloud
- 4:10Code. And now we can see it says GPT
- 4:125.4. If I write, "Hi, can you hear me?"
- 4:15it's now going to be using the GPT 5.4
- 4:17model, and it says, "Yes." Now you know
- 4:19how to change any model you want inside
- 4:21of Cloud Code. But let me show you
- 4:22something that is really, really cool,
- 4:24cuz these models that I've shown right
- 4:25now still does cost something. If you go
- 4:27inside of OpenRouter, you click explore
- 4:29models, it's going to show you all of
- 4:30the models that are inside of here. If
- 4:32you go to prompt pricing, and you just
- 4:34set it to free, it's going to show you
- 4:35all of the free models that are right
- 4:37now inside of OpenRouter. For example,
- 4:39this Qwen 3.6 Plus is one of the best
- 4:42models out there right now, and you can
- 4:43test if it is working by going to the
- 4:45chat and writing, "Hi." See, it says,
- 4:47"Hi there, how can I help you today?" So
- 4:48this model is working. What I'll do is
- 4:50that I'll go back, and then I'm going to
- 4:52copy the model name, go back into our
- 4:55settings {dot} local {dot} JSON, and
- 4:56then change this model. [music] And you
- 4:59can see now it's using the Qwen 3.6 Plus
- 5:02free model, and you can see it says,
- 5:03"Yes, I can hear you. How can I help?"
- 5:05And I don't have to pay anything for
- 5:06using this model. If we go inside of our
- 5:08OpenRouter and inside of logs, we can
- 5:10see how much each request actually cost
- 5:13us. You can see the Gemma 4 model was
- 5:15costing us a little bit, the GPT 5.4
- 5:17models were costing us a little bit, but
- 5:18this Qwen 3.6 Plus is completely free.
- 5:22Is it as good as the paid models? No, of
- 5:24course not. You always get what you pay
- 5:25for, but this is a really good
- 5:27alternative if you hit a rate limit and
- 5:28now you need to use something else. But
- 5:30what do you actually do with this
- 5:32information? When you hit a rate limit
- 5:33all the time, people on Twitter are
- 5:35going crazy about hitting rate limits
- 5:37all the time. just switch to a cheaper
- 5:39model that is still going to perform
- 5:40very, very well. Like if you hit a rate
- 5:42limit on Cloud Code, you can switch to
- 5:44something that is very cheap, like Gemma
- 5:464, or something that is a little
- 5:47cheaper, I think like 50% of the price,
- 5:49like GPT 5.4. If you want to learn more
- 5:52about Cloud Code and more about building
- 5:53an AI agency in general, where you are
- 5:55selling this as a service to businesses,
- 5:57then you can check out our community,
- 5:59The 1% in AI. Inside of here we have
- 6:01every single thing that you need to
- 6:02learn in order to start an AI agency,
- 6:04and we are soon dropping a Cloud Code
- 6:05masterclass, which is hours of content
- 6:07specifically on Cloud Code. And the best
- 6:09thing is that right now we are running
- 6:11the 30-day challenge that gets you from
- 6:13nothing to an AI agency in 30 days. And
- 6:15the best thing is when you stay
- 6:16consistent and you finish the 30-day
- 6:18challenge, we actually get your first
- 6:19month completely refunded, which means
- 6:21if you are committed, you get to join
- 6:23completely for free. If you want to
- 6:25learn more about how that works, then
- 6:26I'll leave a link right below this
- 6:27video. Thank you guys so much for
- 6:29watching. You two things that this video
- 6:30is going to be interesting for you. As
- 6:32always, have an awesome rest of your
- 6:33day. Albert out.
About this transcript
This page contains the full transcript of I fixed Claude Code Usage Limits. Here's how... by Albert Olgaard, generated from the public captions YouTube serves with the video. The transcript has 1,542 words across 206 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.