Although now OpenAI Model prices are getting cheaper, but you still need to pay to use them. If you have no budget, or just want to test or use them for less important things, this article recommends a project shared on GitHub called “Free ChatGPT API.” By simply registering via GitHub, you can get a dedicated API Key for free, allowing you to use OpenAI models without limits. I tested it, and it really works.
How to get unlimited free access to the ChatGPT API
This Free ChatGPT API project is launched by V-API. It’s unclear which company is behind it, but it looks like it’s from China. The API obtained doesn’t directly call OpenAI’s official service but is relayed through free.v36.cm. So chat content may be recorded, analyzed, and used for training. This is why I specifically emphasized earlier that it’s for “testing” or “unimportant things.”
Those who mind these things are advised to skip it.
Additionally, the project also states that the project’s API is intended solely for lawful purposes such as non-commercial learning, research, and scientific testing, and may not be used for any illegal, non-compliant, or commercial purposes.
After clicking the link above to go to the Free ChatGPT API project, scroll down to find the “Free Usage” section, and you will see the “Go to claim free API key” link:
Log in to your GitHub account, and if you don’t have one, just register for free:
Then click Authorize for popjane:
You will receive a dedicated API key and API URL link, which you can then input into the AI service you are using, using the OpenAI format:
Here I, by Cherry Studio For example, Cherry Studio is an all-in-one local AI tool that can not only install local models but also support APIs from various services, including OpenAI. It’s available on Windows and Mac. In the settings, I added a new OpenAI provider type, named it Free, filled in the API key and API URL I just obtained, and then clicked on Model Management below:
Then free-to-use models will appear. Currently available: various versions of GPT-3.5 and GPT-4o-mini:
GPT-4o-mini is the earliest version; click the + next to it to add the model to Cherry Studio:
I added these two: gpt-3.5-turbo-16k and gpt-4o-mini-2024-07-18.
Once you’re in the chat interface, you can select the model you want from the top.
Both models have:
The test chat also succeeded, but the default response is in Simplified Chinese. If you want it to respond in Traditional Chinese, you can emphasize this in the system prompt.
What’s strange, though, is that this gpt-4o-mini model’s knowledge should be up to July 18, 2024, but when I asked the AI, it only replied that its knowledge goes up to October 2023.
Finally, just to remind you again, don’t use this API for anything important.
Source: KOCPC Chinese