Perplexity debuts GLM 5.2 orchestrator with Opus escalation

Perplexity is doubling down on its multi-model routing strategy to balance cost and frontier-level intelligence. Using a post-trained GLM 5.2 as the frontline orchestrator allows for efficient query handling while reserving expensive Opus inference for difficult tasks. Allocating more compute to this component suggests that Perplexity views smart routing as its primary lever for rapid quality improvement, reinforcing their model-agnostic approach and flexibility to integrate the best available technology.