Playground Documentation - PromptAsCode

Overview

The Prompt Playground is a powerful testing environment that allows you to evaluate your prompts across multiple AI models simultaneously. Instead of testing one model at a time, you can compare responses from GPT-4, Claude, Gemini, and other models side-by-side to find the best fit for your use case.

Multi-Model Testing

Test against 10+ models at once

Latency Tracking

Compare response times

Cost Estimation

See API costs per call

How to Use

1
Enter System Prompt (Optional) - Define the AI's behavior, personality, or role. This sets context for how the model should respond.
2
Write Test Message - Enter the user message you want to test. This is what you would typically send to the AI.
3
Select Models - Check the models you want to compare. You can select multiple models to run simultaneously.
4
Adjust Temperature - Set the creativity level (0-2). Lower values give more focused responses, higher values give more creative responses.
5
Click Run - All selected models will process your prompt in parallel and display results side-by-side.

Example System Prompt

You are a helpful coding assistant. You write clean,
efficient code with clear comments. Always explain
your reasoning before providing code.

Available Models

Models are configured by your platform administrator via the Admin Panel. Common available models include:

OpenAI Models

GPT-4o, GPT-4o Mini, GPT-4 Turbo. Excellent at reasoning, coding, and following complex instructions.

Anthropic Models

Claude 3.5 Sonnet, Claude 3 Haiku, Claude 3 Opus. Strong at analysis, creative writing, and nuanced tasks.

Google Models

Gemini Pro, Gemini Flash. Excellent at multimodal tasks and reasoning.

Parameters

Temperature (0-2)

Controls randomness in the output. Lower values make responses more focused and deterministic, while higher values increase creativity and variability.

0-0.3

Factual, consistent

0.4-0.7

Balanced, natural

0.8-2.0

Creative, varied

Understanding Results

Each model response card displays key metrics to help you evaluate performance:

Latency (ms)

Time from request to first response. Lower is better for real-time applications.

Input Tokens

Number of tokens in your prompt. Affects cost and context window usage.

Output Tokens

Number of tokens in the response. Longer responses cost more.

Cost ($)

Estimated API cost for that specific call based on current pricing.

AI Expert Use Cases

Model Selection

AI engineers use the Playground to determine which model best suits their application's needs. By testing identical prompts across models, you can identify which one provides the best quality-to-cost ratio for your specific use case.

Prompt Optimization

Test different prompt variations to see how various models interpret your instructions. This helps identify ambiguities and optimize your prompts for consistent results.

Cost Benchmarking

Compare costs across models for similar quality outputs. Some tasks may be handled equally well by cheaper models, saving significant costs at scale.

Tips & Best Practices

Pro Tips

Be specific in your system prompt about the desired output format
Include examples in your prompt to guide the model's response style
Test the same prompt multiple times to check consistency
Use lower temperature for factual tasks, higher for creative tasks
Copy successful responses to iterate and improve your prompts

Common Mistakes to Avoid

Testing with too high temperature for deterministic tasks
Not providing enough context in the system prompt
Ignoring cost differences when choosing models for production
Using only one test case - always test multiple scenarios