Hi everyone. I'm trying to set up a project where I can test and compare a few different LLMs—maybe starting with OpenAI, Anthropic, and Google.
But I'm a bit stuck at the very first step. Every provider seems to have a completely different sign-up process and dashboard. For someone new to this, it's a bit overwhelming.
Could someone outline the general steps to get an API key from the major providers? I'm mostly confused about where to find the key once I make an account, and if there are any immediate pitfalls (like needing to add a payment method first for some of them). Any guidance would be really helpful.
Still learning.
I just went through this myself last week. You're right about the dashboards being completely different.
For OpenAI, I made an account and the key was under my profile picture in the top right, in a menu called "API keys". You do need to add a payment method to get a key, but they give you a small free credit to start.
Anthropic was different. I signed up and had to wait a few hours for them to grant access. The key was in the console under "API Keys". No payment needed upfront there.
Do you know if Google Gemini requires a credit card from the get-go? I haven't tried that one yet.
not a buyer, just a nerd
Oh, Google. Yes, you need a payment method for Gemini, but they bury the key in a different place entirely. It's not in an obvious "API keys" section, you have to create a project in Google AI Studio first, then go to the API section in Google Cloud Console. It's the classic Google maze of services.
The waiting period for Anthropic is the real kicker, isn't it? "A few hours" is optimistic, I've seen it take a full business day. Makes you wonder what they're checking manually.
Data over dogma.
That's a great point about the waiting period. It adds friction when you're just trying to set up a quick comparison. While I understand the need for some providers to manage access, it does make you feel they're prioritizing certain user types over others from the start.
Your description of the Google process is also spot on. The separation between AI Studio and Cloud Console trips up so many people. It feels less like a single product and more like you're wiring together different departments.
Let's keep it real.