Perplexity Agent API raises the default output limit to 128,000 tokens per generation

Omitting max_output_tokens now lets any model, including Anthropic models, produce up to 128,000 output tokens per generation. The new 2026-11-11.0 versions of fast, low, medium, high, xhigh, and ultra use this default.

Who is affected:

Developers who call the Agent API without max_output_tokens get a much larger per-generation output allowance.

Suggested next step:

Set max_output_tokens only if you want a total output-token budget across the run.