Skip to main content
The Prior Labs API provides secure, cloud-hosted access to TabPFN without managing GPUs or infrastructure. This guide walks you through how to authenticate, upload your data, and make your first API call. The API uses by default the TabPFN-2.5 model. There are options to specify a model checkpoint from the list of available checkpoints on HuggingFace and set it using the model_path parameter. Please refer to the Predict endpoint documentation for detailed instructions.

​
1. Get Your Access Token

  1. Visit ux.priorlabs.ai and sign up/in to your Prior Labs account.
  2. Complete your account setup by entering the required details.
  3. Navigate to Account → Access Tokens.
  4. Click Unmask to reveal your personal API token, then copy it securely.
Your access token grants full API access to your account. Store it securely and never share or commit it to version control.

​
2. Set Up Authentication

You can authenticate and define dataset paths once, and reuse them across both /v1/fit and /v1/predict calls. 
import os, json, requests

# Define your file paths
train_path = "train.csv"   # path to your training dataset
test_path = "test.csv"     # path to your test dataset

# Get your API key from the environment
api_key = os.getenv("PRIORLABS_API_KEY")

headers = {"Authorization": f"Bearer {api_key}"}
We recommend storing your access token in an environment variable rather than directly in code: 
export PRIORLABS_API_KEY="your_api_token_here"

​
3. Train a Model with /v1/fit

Upload your training dataset to the /v1/fit endpoint. The API automatically handles preprocessing and schema inference. 
payload = {"task": "classification"}

files = {
    "data": (None, json.dumps(payload), "application/json"),
    "dataset_file": (train_path, open(train_path, "rb")),
}

fit_response = requests.post(
    "https://api.priorlabs.ai/v1/fit",
    headers=headers,
    files=files,
)

model_id = fit_response.json().get("model_id")
print(f"✅ Model trained: {model_id}")
The API returns a unique model_id - use this ID for subsequent prediction requests.
Dataset Size LimitsThe Prior Labs API currently has the following limits:
  • Per Request Limit: Maximum of 20 million cells per request (e.g., 50K rows × 400 columns)
  • Daily Limit: Total of 100 million cells per day
Plan your data uploads accordingly to stay within these limits.

​
4. Next Steps

Refer to the POST /v1/predict endpoint in the API Reference for detailed request formats, parameters, and response examples.

​
Security Reminder

  • Keep your access token private and rotate it periodically.
  • All requests are fully encrypted in transit (TLS 1.2+).
  • Prior Labs does not store or share your datasets without your consent.
For questions about authentication, access tokens, or API onboarding, contact us at: hello@priorlabs.ai

FAQ

Have questions? Check out the FAQ for answers to common topics about GPUs, limits, API usage, and more.