machined.
Search
K
Comment on page

Configuring OpenAI API Keys

How to configure your API keys
Machined currently operates on a bring-your-own-key basis, so requires an OpenAI API key from a validated paid account with acceptable rate limits. You can follow the steps below to set everything up.

Prerequisites

  1. 1.
    Sign up for an OpenAI Platform account
  2. 2.
    Upgrade your OpenAI Platform account to a paid account
  3. 3.
    Check your rate limits to make sure they are sufficient for the platform
You should ensure that your rate limits are sufficient to use Machined, we require a minimum of 200 RPM and 50,000 TPM limits for the gpt-3.5-turbo and gpt-3.5-turbo-16k models.
Your ChatGPT account is a completely different service and is not compatible.

Configuration

In Machined, open the Configuration popup from the dashboard.
Opening the configuration settings
Click on the OpenAi Configuration section:
Paste your API Key into the input and click Verify:
Machined will verify and validate your key, if successful you will see a redacted key with a green tick. You will also see your API rate limits summarized for the relevant models.
If instead you see an error, please consult our troubleshooting guide below.

Troubleshooting

If you receive an error when setting your key, this usually means that there is a problem with either your OpenAI account or API key. Common reasons for invalid keys:
  • Typos - mistyped or incorrectly copied key
    • Copy the key again, ensuring it is correct
  • Throttling - an intermittent error with your api account
    • Wait a few minutes and try again
  • Revoked key - you are using an old key that has been revoked
    • Create a new key in your account and try again
  • You are on a Free OpenAI account - free accounts are not supported
    • Add a payment method and enable billing on your OpenAI Platform account
  • You have an insufficient Rate Limit - you need an account with sufficient rate limits
    • Double check your rate limits and ensure they are sufficient
    • You could reach out to OpenAI and ask for a rate limit increase