Prime Intellect Evals SDK

0.2.0 verified Sat May 09 auth: no python

Prime Intellect Evals SDK (v0.2.0) for pushing and managing evaluations. Supports Python >=3.10 and is used to submit AI evaluation results to the Prime Intellect platform. Released weekly or monthly.

pip install prime-evals

Common errors

error ModuleNotFoundError: No module named 'prime_evals' ↓

cause Installed 'prime-evals' package but tried to import 'prime_evals' incorrectly.

fix

Run: pip install prime-evals Then use: from prime_evals import ...

error prime_evals.exceptions.AuthenticationError: Invalid API key ↓

cause Missing or incorrect API key passed to PrimeClient.

fix

Obtain a valid API key from Prime Intellect and set it as: client = PrimeClient(api_key='your-key')

Warnings

gotcha The package is named 'prime-evals' on PyPI but imports as 'prime_evals' (underscore). Do not use 'import prime_evals' without installing the correct package. ↓

fix Install with 'pip install prime-evals' and import as 'from prime_evals import ...'.

gotcha API key must be passed explicitly; environment variable 'PRIME_API_KEY' is not automatically read. ↓

fix Always pass api_key as argument: PrimeClient(api_key='your_key') or read from env manually.

Imports

PrimeClient
```
from prime_evals import PrimeClient
```
Official import path as per docs

Quickstart

Initialize client and list supported evaluation types.

import os
from prime_evals import PrimeClient

api_key = os.environ.get('PRIME_API_KEY', '')
client = PrimeClient(api_key=api_key)
print(client.supported_evals())