Running & Comparing

Run prompts in real time with streaming output, and compare up to four provider/model targets side-by-side.

Real-time streaming

Runs stream back token by token over Server-Sent Events, so you see output as the model produces it. Each run reports the model used, token counts, and an estimated cost when the stream completes.

Comparing models

The Test tab lets you place up to four provider/model cards side by side and run them against the same variables at once — perfect for choosing between, say, GPT-4o, Claude, and Gemini, or comparing temperatures. Your compare layout is saved, so it's there when you return.

• Each card picks its own provider, model, and parameters.
• Run all cards together and read the streamed outputs in parallel.
• Great for picking the cheapest model that still passes your evals.