PerplexityModel Selection Dropdown
Perplexity offers a comprehensive model switcher directly in the composer, displaying a wide array of AI models including GPT-5.2, Claude Opus 4.5, Gemini 3 Pro/Flash, Grok 4.1, and proprietary Sonar models.
Model picker with Sonar, GPT, Gemini, Claude, and Kimi options in Search mode
What's happening
Perplexity distinguishes itself by offering the widest selection of underlying AI models directly in the composer. Users can choose from multiple providers (OpenAI, Anthropic, Google, xAI, and proprietary models), with clear visual indicators for premium ("max") and recently added ("new") options.
Patterns
Dropdown menu displaying diverse model options with contextual badges and descriptions
UX Insights
- •Offering multi-provider model choice positions Perplexity as a model-agnostic platform
- •Visual badges ("max", "new") help users identify premium or latest features
- •Placing selection in composer rather than settings emphasizes per-query flexibility
- •Including regional hosting info (e.g., "Hosted in the US") addresses privacy concerns
- •Default "Best" option simplifies choice for non-technical users
Design Decisions
Perplexity embraces a multi-model strategy by surfacing model selection prominently in the composer. This design positions the platform as a meta-layer that lets users access the best model for each query, rather than locking them into a single provider's ecosystem. The "Best" default option maintains simplicity while the full dropdown empowers power users.
More case studies
ClaudeModel Switcher with Hierarchy
Claude implements a tiered model selection system that balances simplicity with power user needs. The main dropdown presents three current models with clear use-case descriptions, while a "More models" submenu provides access to legacy versions.
DeepSeekTool Switching in Composer
DeepSeek composer: Instant/Expert tier toggle, DeepThink and Search chips in the card, and attach with tier-specific limits.
Google GeminiMode-Based Selection
Gemini reframes model selection as "mode" selection, presenting users with three performance tiers, Fast, Thinking, and Pro, each optimized for different task types and response complexity.
GrokModel Selection
Grok offers intuitive model selection with Auto mode that intelligently chooses between Fast and Expert models, plus additional specialized models including Grok 4.1 Thinking, Heavy, and premium SuperGrok.