Perplexity AI Introduces Four-Tier Effort Settings for Task Optimization

Perplexity has announced a significant update to its Computer product, introducing a granular "effort" adjustment system designed to optimize computational resources. Users can now navigate four distinct levels of reasoning depth via a dedicated slider, allowing the platform to tailor its processing power to the complexity of specific queries. This functionality aims to improve efficiency in task orchestration, ranging from basic data synthesis to complex market analysis.

Scalable Reasoning for Complex Data Processing

The new adjustment feature allows users to select between Light, Standard, High, and Ultra modes. According to the Perplexity development blog, these settings determine the sophistication of the primary model responsible for task orchestration. Once a level is selected, the orchestrator delegates specific sub-tasks to a network of auxiliary agents, which may utilize diverse architectures from various LLM (Large Language Model) vendors.

• Light: Optimized for routine administrative tasks, such as summarizing vendor invoices or extracting data from simple documents.
• Standard: Balanced for general inquiries requiring moderate synthesis.
• High: Designed for multi-step workflows requiring deeper technical insight.
• Ultra: Reserved for open-ended, high-level strategic questions, such as evaluating new markets or cross-referencing vast datasets.

Resource Management and Multi-Platform Rollout

A critical component of this update is the integration of a dynamic credit point consumption model. Instead of a flat fee, the platform now deducts credits based on the actual workload and the reasoning depth selected by the user. Lower levels will consume fewer resources, providing a cost-effective solution for high-volume, low-complexity operations. While the "effort" slider provides a new layer of control, Perplexity confirmed that custom model selection remains accessible for power users who prefer specific AI architectures.

The feature is currently live for users on the web interface, as of September 18, 2026. The company has indicated that mobile applications for Android and iOS will receive the update in the near future. For the cryptocurrency and blockchain sector, where rapid data analysis and on-chain research are vital, such tools represent an evolution in how decentralized information is processed and interpreted by artificial intelligence.

"The orchestrating model assigns work to auxiliary agents... Credit point consumption depends on the actual workload, with lower levels consuming less."

The implementation of these tiers reflects a broader trend in the AI industry toward agentic workflows and resource efficiency. By allowing users to toggle the "reasoning intensity", Perplexity is positioning its tool as a versatile assistant capable of handling both mundane data entry and high-stakes financial research. As AI continues to integrate with Web3 ecosystems, the ability to scale computational effort will likely become a standard requirement for professional analytical platforms.