Granite Playground

Learn aboutGranite 4.2 Language Models

Explore trade-offs related to thinking and tool calling.

  • Using the thinking mode typically yields better results at the cost of longer runtime and a higher token count.
  • If you ask non-trivial questions, the application will give you an answer by an assistant using the regular (non-thinking) mode, the thinking mode with effort set to low, and the thinking mode.
  • It shows the impact of tool calls, search, in this case, the depth and currency of the result.
Using the thinking mode typically yields better results at the cost of longer runtime and a higher token count.

Try up to 3 turns of prompts, or start with the examples below.