Perplexity AI Available Models: All Supported Models, Version Differences, Capabilities Comparison, And Access Requirements

Perplexity AI delivers a diverse model ecosystem for both consumer users and developers, combining an in-app model selector with an API-exclusive Sonar family. Model access, capabilities, and performance differ by surface and subscription plan, shaping user experience across research, reasoning, and advanced retrieval tasks.

·····

The In-App Model Catalog Includes Major AI Models For Pro Search And Max Plans.

Perplexity’s web and mobile app enables users to select from a curated lineup of leading language models for Pro Search mode, while an auto-select “Best” mode chooses the optimal model for standard queries. Users with Pro or Max subscriptions gain access to a broader set of models and features.

Pro Search exposes Sonar, GPT-5.2, Claude Sonnet 4.5, Gemini 3 Pro, Grok 4.1, and Kimi K2 Thinking. The highest Max plan additionally unlocks Claude 4.5 Opus, a model reserved for demanding reasoning and agentic use cases. Each model supports a reasoning toggle, enabling or disabling advanced logical processing.

The Best mode delegates model selection to the system, optimizing for speed and accuracy without fixed model assignment. Model availability and selection may also depend on subscription level and region.

........

Perplexity In-App Model Options And Access Tiers

Model Name	Highlighted Strength	Access Requirement
Sonar	Fast web search, grounded summarization	Pro Search (default)
GPT-5.2	Reasoning, coding, creativity	Pro Search
Claude Sonnet 4.5	Technical reasoning, coding	Pro Search
Gemini 3 Pro	Multimodal, code generation	Pro Search
Grok 4.1	Conversational search, up-to-date answers	Pro Search
Kimi K2 Thinking	Problem solving, logic	Pro Search
Claude 4.5 Opus	Deep reasoning, agentic tasks	Max plan only

Subscription level determines model access in the app.

·····

The Perplexity API Exposes The Sonar Model Family For Search, Reasoning, And Research.

The developer-facing Perplexity API features the proprietary Sonar model family, with models specialized for search, multi-step reasoning, and exhaustive research. API users specify the model in each request, selecting from sonar, sonar-pro, sonar-reasoning-pro, and sonar-deep-research.

Sonar is optimized for lightweight, cost-effective search and summarization, while Sonar Pro offers deeper context and broader retrieval, including 2x more search results and a larger context window. Sonar Reasoning Pro excels at chain-of-thought and complex analytical tasks. Sonar Deep Research is designed for in-depth, long-form synthesis, handling hundreds of sources for detailed reporting.

The API context lengths are substantial: Sonar and Sonar Reasoning Pro support 128,000 tokens, Sonar Pro extends to 200,000 tokens, and all models feature strict separation from customer data for privacy.

........

Perplexity API Sonar Models And Version Differences

Model	Category	Context Length	Primary Focus
sonar	Search	128,000 tokens	Fast, cost-efficient Q&A, summaries
sonar-pro	Search	200,000 tokens	Deep retrieval, complex queries
sonar-reasoning-pro	Reasoning	128,000 tokens	Multi-step logic, chain-of-thought
sonar-deep-research	Research	128,000 tokens	Long-form, exhaustive synthesis

Models are chosen based on the depth and type of analysis required.

·····

Capability Differences Are Driven By Search Depth, Reasoning, And Context Budget.

Perplexity’s model catalog is structured for specific workflows. Search models like Sonar and Sonar Pro excel at rapid Q&A and content summarization, with Sonar Pro offering a broader context window and access to more search results.

Reasoning models, such as Sonar Reasoning Pro, are engineered for multi-step analysis, structured problem solving, and chain-of-thought outputs. Research models like Sonar Deep Research synthesize information across a wide range of sources for detailed, report-style outputs.

In the app, model differences reflect their training and optimization—some are designed for speed, others for multimodal input or high-level reasoning. Higher context windows in Sonar Pro enable longer, more complex prompts, benefiting users with intensive research or document analysis needs.

........

Perplexity Model Capabilities Comparison

Capability	Sonar	Sonar Pro	Reasoning Pro	Deep Research
Speed	High	Moderate	Moderate	Lower
Retrieval depth	Standard	2x more	Standard	Maximal
Reasoning	Basic	Enhanced	Strongest	Strong
Context length	128K	200K	128K	128K
Output style	Summaries, Q&A	Detailed, broad	Stepwise logic	Report synthesis

Each model excels in its targeted application.

·····

Access Requirements Depend On Plan, Platform, And Intended Use.

App access to advanced models and features is governed by subscription plans. The Pro plan unlocks additional models and Pro Search mode, while the Max plan grants access to the most powerful models such as Claude 4.5 Opus. Certain models, features, or context budgets may be region-restricted or phased in by rollout.

API access requires developer registration, obtaining an API key, and selecting supported Sonar models in requests. API usage is governed by Perplexity’s billing structure and does not offer in-app model choices such as GPT-5.2 or Gemini Pro.

For enterprise users, data handling policies change: Enterprise plans default to AI data retention off and do not include customer data in model training, while API model documentation highlights a “No training on customer data” promise across the Sonar family.

........

Perplexity Model Access Requirements And Plans

Surface	Models Available	Access Mechanism	Subscription/Plan
In-app Pro Search	Sonar, GPT-5.2, Claude 4.5, Gemini 3 Pro, Grok, Kimi K2	App login, model picker	Pro/Max required for full access
In-app Max features	Claude 4.5 Opus	Max subscription	Max only
API (developer)	Sonar family	API key, request param	Developer account
Enterprise	Sonar (API), custom data handling	Admin controls	Enterprise plan

Plan and platform shape available features and models.

·····

Perplexity AI’s Model Ecosystem Supports Search, Reasoning, And Research Across Plans And APIs.

Perplexity AI combines curated access to major language models in the app with a specialized, privacy-focused Sonar model suite for developers. Subscription level, platform, and use case determine which models and features are available, ensuring flexibility for research, retrieval, and reasoning tasks at scale.

·····

FOLLOW US FOR MORE.

·····

DATA STUDIOS

·····

[datastudios.org]

·····