Use ChatGPT with API Key to save money
GPT-4 is good but If you're someone who just use it rarely you wouldn't like the idea of paying $20 for it each month. If you're that person then I hae a solution for you! OpenAI provide API access to GPT-4 and it is billed only for what you use. So if you don't us it you won't get charged at all! But to use the API you need an interface that works with the API.
Introducing PulsarChat
PulsarChat works with your OpenAI API key, so you only pay to OpenAI for what you use. For an average user, this is 10x cheaper than paying $20 every month with ChatGPT Plus subscription to have limited access to GPT4 (they charge even if you didn’t use it!).
With PulsarChat you can use GPT4 all day with no restrictions! If you don't have GPT4 access in your API Key, you can unlock it once you pay at least 1 OpenAI bill.
How it works
- Get your OpenAI API Key (If you don't know how, stick till the end.)
- Enter the key in PulsarChat
- Start Chatting!
You can use PulsarChat with 2 service providers. One is obviously OpenAI and the other is OpenRouter.
Currently, the most powerful model out there is GPT-4, and it is available through both OpenAI and OpenRouter endpoints. But if this is the first time you're going to use your OpenAI API key then you have to wait at least 1 month and must have paid at least one bill to OpenAI to be eligible to use GPT-4.
That is why we support OpenRouter too, which allows you to access GPT-4 without any restrictions just like it is with OpenAI, but without the need to wait for 1 month to be eligible. And, OpenRouter supports some extra models like Mistral 7B and more...
Here are the costs (same for both API providers)
| Model | Input Cost | Output Cost |
|---|---|---|
| GPT-4 (8K Context) | $0.03 / 1K tokens | $0.06 / 1K tokens |
| GPT-4 (32K Context) | $0.06 / 1K tokens | $0.12 / 1K tokens |
| GPT-3.5 (4K Context) | $0.0015 / 1K tokens | $0.002 / 1K tokens |
| GPT-3.5 (16K Context) | $0.003 / 1K tokens | $0.004 / 1K tokens |
In terms of token cost per one US dollar,
| Model | Input Cost | Output Cost |
|---|---|---|
| GPT-4 (8K Context) | 33.33K tokens | 16.67K tokens |
| GPT-4 (32K Context) | 16.67K tokens | 8.33K tokens |
| GPT-3.5 (4K Context) | 666.67K tokens | 500K tokens |
| GPT-3.5 (16K Context) | 333.33K tokens | 250K tokens |
For other models that are available through OpenRouter see the rates here.
But currently, OpenRouter API doesn't support Plugins, so there is that. If you've decided what API Key you're going to use. Read one of the posts (or both if you're interested in using both API Keys) below to get started.