Perplexity cuts per-query cost of every Agent API effort level and adds an ultra tier

Perplexity announced on November 5, 2026 that every effort level in its Agent API now costs less per query. The company also added a new ultra effort level aimed at broad research.

What changed

According to Perplexity, research quality is higher or unchanged across the presets after the cost reductions.

The company's headline example is the medium level:

• Medium improved from 55% to 77% on BrowseComp and DSQA, according to Perplexity.
• Its cost fell almost 90%, the company says.

The new ultra level is available for broad research.

How effort levels work

Effort levels are named presets in the Agent API. Each one packages model choice, research-step and tool-call limits, search behavior and available tools. Developers use them to select a balance of cost and speed without configuring each of those parameters separately.

What this means

The results come from Perplexity's own reporting, so the benchmark gains and cost reductions are company claims. If they hold in developers' own workloads, the medium preset would be cheaper per query and score higher on the two benchmarks Perplexity cites. The ultra level extends the preset range for broad research tasks. Because effort levels bundle several settings under one name, developers can adopt the new pricing and the new tier by choosing a preset rather than reworking their configuration.