Perplexity Cuts Agent API Research Costs and Adds an Ultra Effort Level

At a glance

Perplexity announced that it has lowered the cost per query for its Agent API at every reasoning level. At the same time, it launched a new Ultra level for broad research. Perplexity says research quality has stayed the same or improved even though costs have dropped.

Key points

1. 1

   You can choose a reasoning level (a preset bundle of settings for the model, number of steps, and tool-use limits) to balance cost, speed, and research depth.

2. 2

   Ultra splits a problem among three agents that each investigate it, and a fourth agent combines their results.

3. 3

   Limits on the number of steps and tool calls prevent overspending. The system also shows the remaining budget so it can finish its answer.

4. 4

   A new feature shortens search results and page content, so less unnecessary information is used.

Why it matters

As the cost of AI gathering many sources and writing answers falls, it can be used in more places. It also changes how developers choose between cost and quality.