Skip to main content

​
Playground

The LLMGrid Playground is an interactive web interface that lets you experiment with prompts, compare responses across multiple LLMs, and visualize routing strategies—all without writing code.

​
Key Features

  • Prompt Testing
    Enter prompts and view responses from different models side-by-side.
  • Multi-Model Comparison
    Compare latency, cost, and output quality across providers (OpenAI, Anthropic, Azure OpenAI, Google Gemini, etc.).
  • Routing Simulation
    Preview how your configured routing strategy (cost-optimized, latency-optimized, balanced) selects models.
  • Streaming Support
    See token-by-token generation in real time for streaming-enabled models.
  • Usage Metrics
    Track token counts, estimated cost, and latency for each request.

​
How to Access

  • Navigate to Playground in the LLMGrid Console:
    https://app.llmgrid.ai/playground
  • Select:
    • Tenant and Project
    • Route or Direct Model
    • Optional: Adjust temperature, max tokens, and other parameters.

​
Example Workflow

  1. Choose Route: chat_default
  2. Enter prompt:
  3. Click Run to view:
  • Responses from all models in the route
  • Latency and token usage per model
  • Routing decision logic applied

​
Why Use Playground?

  • Validate prompt templates before deploying to production.
  • Benchmark models for cost and performance.
  • Debug routing policies and fallback behavior.
Tip: The Playground uses the same OpenAI-compatible API as your apps, so results mirror real API calls.