Transcription
[music] Hi, welcome to another video.
So, there is a pretty crazy free model available inside Open Code right now. It is called Big Pickle, which obviously tells you absolutely nothing about the model. Open code officially describes Big Pickle as a stealth model, but I can confirm that the model currently running behind it is GLM 5.2. Yes, the same GLM 5.2 that is one of the strongest open-weight coding models right now can currently be used for free through Open Code, and the limits are actually quite good.
Open Code currently shows a limit of around 200 requests per 5-hour window for Big Pickle and its other free models. People using it generally say that this can translate into roughly 5 hours of coding work, depending on what you're doing. Now, that does not mean there is a stopwatch that gives you exactly 5 hours of non-stop usage. The limit is based on requests. A simple question may use one request, while an agentic task that reads files, runs commands, edits code, checks the result, and fixes problems can make many model requests in the background. So, if you ask it to build a complete application and then disappear for lunch, you may hit the limit faster. But for normal interactive coding, repo exploration, bug fixes, refactoring, and smaller features, 200 requests is actually pretty generous for something that costs nothing.
There are also reports of this setup running at more than 280 tokens per second with a time to first token below 0.8 seconds. Obviously, speed will depend on demand, location, and how Open Code is serving it at that moment, but it feels extremely fast when the endpoint is not overloaded. So, this is kind of awesome. You're getting a genuinely capable open-weight coding model inside one of the best open-source coding harnesses with a free limit that is large enough to do real work.
Now, let me show you how to use it. First, you need to install Open Code. On macOS, Linux, or Windows through WSL, you can open your terminal and run `curl -fsSL https://openco.ai/install | bash`. If you already have Node.js installed, you can instead run `npm install -g openode-ai`.
Once Open Code is installed, go into the project folder that you want to work on and run `open code`. If you have not connected Open Code Zen before, run the `/connect` command. Select Open Code Zen, follow the authentication link, sign in, create an API key, and paste that key back into Open Code. Then run `/models` and search for Big Pickle. This part is important. Select the model called Big Pickle if you want the free version. You may also see a separate model explicitly called GLM 5.2. That is the normal paid GLM 5.2 endpoint. Big Pickle is the free stealth model route. So, do not accidentally select the paid one and then wonder why it is using your Zen balance.
Once Big Pickle is selected, you can start using it like any other model in Open Code. For example, you can say, "Inspect this repository, explain how the project is structured, and tell me where the authentication logic is implemented." Or you can say, "Find the cause of this bug, make the smallest safe fix, run the relevant tests, and then summarize what you changed."
For a larger task, I would recommend starting in plan mode. Ask it to inspect the codebase, identify the files that need to change, and give you a clear implementation plan. Once the plan looks correct, switch to build mode and ask it to implement it. This generally works better than immediately telling any coding agent to rewrite half of your project. GLM 5.2 is good, but it is still an AI model, and giving it a clear scope makes a massive difference. I would also tell it to keep changes focused, avoid unrelated refactoring, run the existing tests, and show you the final diff.
A good prompt would be something like this: "First, inspect the relevant files and explain the current implementation. Then add this feature while keeping the changes scoped. Run the tests and type checks. Fix any issues you introduced and summarize every file that you changed." That kind of prompt gives the model a process instead of just throwing a vague request at it.
Now, there are two catches that you need to know about. First, this free access is officially described as being available for a limited time. Open Code can change the model, reduce the limits, or end the free period later. Big Pickle is a stealth alias, so I would not assume that the name will always point to GLM 5.2 either. As I am making this video, it is GLM 5.2 and it is free, but treat this as something to use while it is available, not as a permanent promise.
Second, Open Code says that data collected while using Big Pickle during the free period may be used to improve the model. So, do not use the free endpoint with private company code, credentials, customer data, confidential documents, or anything else that you would not want included in model improvement data. Use it on open-source projects, personal experiments, test repositories, and non-sensitive work. If privacy matters, use a provider with the data retention policy that your project requires.
If you hit the free limit, you can wait for the 5-hour window to reset, switch to one of Open Code's other free models, use your own provider, or move to a paid option. Open Code also has its Go plan, which gives the named GLM 5.2 model much larger limits. But honestly, I would try Big Pickle first. 200 free requests per 5 hours is enough to find out whether GLM 5.2 actually works well for your coding workflow before paying for anything.
And from my own GLM 5.2 testing, this is not just some weak model being offered for free because nobody wants it. It is genuinely very capable. It is good at agentic coding. It follows longer tasks well. It works nicely with Open Code, and the current serving speed makes it feel even better. It will not replace Opus or CodeX for every single task, and you should still review the code it writes, but as a free coding option, this is extremely difficult to complain about. GLM 5.2 is currently hiding behind a pickle. Open Code is giving you around 200 requests every 5 hours, and you can be using it in a few minutes. Chef's kiss. Really good stuff. Overall, it's pretty cool.
Anyway, let me know your thoughts in the comments. If you like this video, [music] consider donating through the Super Thanks option or becoming a member by clicking the Join button. Also, [music] give this video a thumbs up and subscribe to my channel. I'll see you in the next one. Until then, bye. >> [music]