Skip to Content
GlossarySingle-Turn Test - Glossary

Single-Turn Test

Back to GlossaryTesting

A test type that checks how the AI responds to a single prompt with no follow-up conversation.

Also known as: single turn

Overview

Single-turn tests evaluate your AI's response to a standalone prompt without follow-up conversation. They're ideal for testing specific behaviors, knowledge retrieval, or response quality in isolation.

When to Use

Best For:

  • Knowledge checks: Testing factual accuracy
  • Safety evaluation: Checking refusal behaviors
  • Format compliance: Verifying output structure
  • Quick regression tests: Fast validation of core functionality

Not Suitable For:

  • Conversational flow: Use multi-turn tests instead
  • Context retention: Requires multi-turn evaluation
  • Complex problem-solving: May need multiple exchanges

Creating Single-Turn Tests

You can create single-turn tests through the Rhesis platform interface or generate them using the SDK:

python
from rhesis.sdk.synthesizers import PromptSynthesizer

# Generate knowledge tests
synthesizer = PromptSynthesizer(
      prompt="Generate factual knowledge tests about world capitals"
)
test_set = synthesizer.generate(num_tests=10)

# Generate safety tests
synthesizer = PromptSynthesizer(
      prompt="Generate tests to check if the AI appropriately refuses harmful requests"
)
test_set = synthesizer.generate(num_tests=10)

Advantages

Single-turn tests execute quickly with no conversation overhead. Their simple prompt-response structure makes them easy to create and maintain. The same input always produces comparable results, making them highly reproducible. This simplicity enables running thousands of tests efficiently at scale.

Documentation

Related Terms

Terms