Quickstart

At this time, High Dimensional Research products are intended for developers and a more technical audience.

In order to make use of High Dimensional Research APIs, you will require the following:

  • access to a language model API or endpoint, including an API key for a service of your choice and/or an endpoint for your model
  • a development environment that can make HTTP requests, for example, using cURL on the command line; fetch() in JavaScript, or the requests module in Python
  • an account with High Dimensional Research, and a generated API key for your organisation

Language model endpoints

Some of the most common language models available at time of writing include the GPT series by OpenAI and Claude by Anthropic. To generate an API key for these models, you can consult the account settings with OpenAI or Anthropic.

Ensure you specify the provider of the model and the model in question when using the High Dimensional Research API. For example, to use GPT-4:

"modelArgs":{
"apiKey": "YOUR-KEY",
"provider": "openai",
"model":"gpt-4-1106-preview"
}

Similarly, you can pass anthropic and claude-2.1 as the provider and model.

If you are running your own model, you can simply pass the endpoint to the High Dimensional Research API with the custom provider.

API key generation

When you gain access to the High Dimensional Research dashboard, visit your API settings and click GENERATE NEW KEY. This key will only be shown once, so ensure you retain it in a safe place. You can add or delete any number of keys at any time.

High Dimensional Research uses a computing credit system. You can check how many credits you have at the Add Credits page.

Perform your first request

Let's start with a simple use of the High Dimensional Research API: a deeply configurable internet search with your preferred model.

  • cURL
  • Python
curl -X POST https://api.hdr.is/browse \
-H "Content-Type: application/json" \
-H "Authorization: Bearer hdr-***" \
-d '{
"objective": "how many people have edited wikipedia",
"startUrl": "http://google.com",
"maxIterations": 10,
"stream": true,
"modelArgs":{
"apiKey": "OUR-KEY",
"provider": "openai",
"model":"gpt-4-1106-preview"
}
}'

In the above example:

  • We've passed our High Dimensional Research key in the Authorization header.
  • We have our objective: "how many people have edited Wikipedia?"
  • By setting maxIterations to 10 we are limiting the amount of pages within the browse session.
  • By setting stream to true we can see each step of the browse session.
  • Finally, modelArgs configures the model that will operate the browser.

Let's see what we get!

{
"id": "e4251f48-a29f-4eb8-9db6-dcc40c8f3045",
"data": {
"commands": [
"Type"
],
"description": "To find out how many people have edited Wikipedia, I searched `how many people have edited Wikipedia` using the combobox at index 8."
},
"object": "stream.command",
"stop_reason": null
}
{
"id": "e4251f48-a29f-4eb8-9db6-dcc40c8f3045",
"data": {
"commands": [
"Type"
],
"description": "Searched for the `number of Wikipedia editors` to find out how many people have edited Wikipedia."
},
"object": "stream.command",
"stop_reason": null
}
{
"id": "e4251f48-a29f-4eb8-9db6-dcc40c8f3045",
"data": {
"commands": [
"Type"
],
"description": "Found the total number of registered users and the number of active editors on Wikipedia within the last 30 days as per the Featured snippet from the web."
},
"object": "stream.command",
"stop_reason": null
}
{
"id": "e4251f48-a29f-4eb8-9db6-dcc40c8f3045",
"data": {
"stopUrl": "https://www.google.com/search?q=number+of+Wikipedia+editors&sca_esv=593784007&ei=sfiKZZOtBIe-kPIPwaKw0As&ved=0ahUKEwjTye_JvK2DAxUHH0QIHUERDLoQ4dUDCBA&uact=5&oq=number+of+Wikipedia+editors&gs_lp=Egxnd3Mtd2l6LXNlcnAiG251bWJlciBvZiBXaWtpcGVkaWEgZWRpdG9yczIFEAAYgAQyCxAAGIAEGIoFGIYDMgsQABiABBiKBRiGAzILEAAYgAQYigUYhgNIpwtQPVjpCnABeAGQAQCYAWWgAcsNqgEEMjUuMbgBA8gBAPgBAcICChAAGEcY1gQYsAPCAhEQLhiABBiKBRiRAhixAxiDAcICCxAAGIAEGIoFGJECwgILEC4YgAQYsQMYgwHCAg4QLhiABBiKBRixAxiDAcICCxAAGIAEGLEDGIMBwgIREC4YgAQYsQMYgwEYxwEY0QPCAiAQLhiABBiKBRiRAhixAxiDARiXBRjcBBjeBBjgBNgBAcICEBAAGIAEGIoFGJECGEYY-wHCAgoQABiABBiKBRhDwgIQEAAYgAQYigUYQxixAxiDAcICDhAAGIAEGIoFGLEDGIMBwgIcEAAYgAQYigUYkQIYRhj7ARiXBRiMBRjdBNgBAsICCBAAGIAEGLEDwgIREAAYgAQYigUYkQIYsQMYgwHCAgcQLhiABBgKwgIOEC4YgAQYxwEYrwEYjgXCAgYQABgWGB7CAggQABgWGB4YD-IDBBgAIEGIBgGQBge6BgYIARABGBS6BgYIAhABGBM&sclient=gws-wiz-serp",
"result": "According to the search results, 46,675,005 users have registered a username on the English Wikipedia, and 120,734 of these users have edited within the last 30 days.",
"pageContent": "..."
},
"object": "stream.result",
"stop_reason": "objectiveComplete"
}

It looks like we've successfully managed to achieve our objective! Of course, if we didn't pass stream: true, we would've simply gotten the response directly.

With this in mind, thank you for reading this guide! We at High Dimensional Research look forward to seeing how you manage to use our products and APIs. Feel free to contact us directly using our company directory with any feedback.